Re: What happened to resumable instructions?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] writes: I just noticed this thread today. I see there is a lot of speculation and incorrect statements, so let me try to clarify. First, from a processor implementation point of view, there is no problem with the old-style MVCL/CLCL interruptible instructions. At appropriate points based on pending interrupts and potential page crossings leading to an access exception, the millicode exits from their execution and sets the GRs and PSW appropriately to continue later. There is no risk of checkstops and it is perfectly well architected to handle a page fault in the middle, as some have speculated. The reason for the new CC3-style interruptible instructions is very simple. It was requested by software developers (both internal and external to IBM). It allows more flexibility in handling other system activity that is not interrupt driven. So for example, software can go off and perform some housekeeping while a long running MVCLE is executing. Note that POPS requires the processor to exit with a CC3 every (approximately) 4KB processed. So for a multi-page move, the overhead of starting and stopping can actually slow down the throughput of the move very slightly. I would expect that all future interruptible ops be of the CC3-style. That said, there will be at least one new interruptible version of an existing instruction announced soon, however, it is an instruction never used by the vast majority of software developers. re: http://www.garlic.com/~lynn/2008c.html#67 What happened to resumable instructions? long storage-to-storage operations with lots of accesses back to real storage ... is costing ever increasing number of processor cycles (as mismatch between processor speed and memory speed increases) ... recent reference http://www.garlic.com/~lynn/2008c.html#92 CPU time differences for the same job one could claim that nearly the same effective results (cc3-style) could be achieved for old-style interruptable instructions under program control ... but requiring a few more registers for loop control ... i.e. actual length is kept in other registers and the lengths used for mvcl/clcl instructions being limited (to 4096). there are other thruput issues associated with long storage-to-storage operations; even back to working on original mainframe tcp/ip product implementation. at the time, some of the competitive tcp/ip implementations were looking at 5k pathlength and five buffer copies ... and there was comparison with something like 150k instruction pathlength and 16 buffer copies for lu6.2. at that time, assuming 8kbyte NFS size buffer ... the processor overhead for the 16 (LU6.2) 8kbyte storage-to-storage buffer copy operations exceeded the processor time for the rest of the pathlength. the other factor that some processors provided was cache-bypass, storage-to-storage instructions. Large number of significant sized storage-to-storage operations ... not only has ever increasing significant overhead in terms of processor cycles ... but can have an extremely detrimental effect on cache occupancy (the actual data in most of the buffers has little probability of ever being needed in the cache, but replacing data that would be needed). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: CPU time differences for the same job
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. Neal Eckhardt [EMAIL PROTECTED] writes: We also had the batch window running so long that we were missing deadlines, and the CIO asked me to review the program stream to see what could be done to make it run faster (yea, good luck with that, they were all database programs and that's a different group). I told him that our new CPU (of essentially the same speed) that was due to arrive in two weeks had two processors rather than the four processors our current machine had, and the problem would probably be resolved with the installation of our new CPU. re: http://www.garlic.com/~lynn/2008c.html#88 CPU time differences for the same job the issue raised in post about disk thruput ... was over a period of years, processors got around 50 times faster in a period that disks only got 3-4 (maybe 5) times faster. Some claims are that the relative abundance of processor power led to a lot of (processor) implementation inefficiencies ... which could be relatively straight forwardly improved if anybody were to pay any attention. as circuit size decreased ... and processor thruput increased ... lots of signal latencies started to play larger and larger factor. the latencies (measured in number of processor cycles) that played a part in processor/disk thruput ... started to also dominant processor real storage ... and in much the same way that real storage started to be used as caches to compensate for disk thruput ... processor caches became more more important compensating for memory latency. in fact, using processor cycles as a measure of memory access time ... became one way of easily recognizing the problem (memories got faster, but processors got much, much faster) ... especially when references were to cache-miss and memory latency measured in possibly thousands of processor cycles. for little drift, recent posts mentioning improving application hitting (overnight) batch window limits http://www.garlic.com/~lynn/2008b.html#74 Too much change opens up financial fault lines http://www.garlic.com/~lynn/2008c.html#24 Job add for z/OS systems programmer trainee and for something totally different ... recent discussion of electronic commerce implementations running into problems hitting overnight window limit http://www.garlic.com/~lynn/2008c.html#85 Human error tops the list of security threats multitasking and multithreading have been used in the past attempting to mask disk access latencies ... keeping processor busy while something was wait for disk access. earlier in this decade ... there were chip hyperthreading solutions also looking at keeping processor functional units busy ... compensate/mask cache-miss memory access latency. the number of processor functional units weren't actually increased ... but there was hardware emulation of two processors ... basically two instruction streams, two sets of registers, etc ... but actual execution being done by common functional units. possibly the original in this genre was a project to do a dual i-stream implementation for 370/195 (i.e. emulation of two-processor smp) ... which never actually shipped as product. the issue was that 195 had a 64 instruction pipeline that could execute instructions concurrently with common set of functional units. the pipeline could go a long way towards masking the difference in processor thruput and memory latency (w/o actually having a cache). The problem was that the amount of processor logic/memory supporting this function was rather limited. These days, chip implementations may have several hundred positions for dealing with branch prediction and speculative execution (and backing out instructions not actually branched to). In the 370/195, except in very special cases, branches would drain the pipeline. Normal codes, with typical branch useage, resulted in 195 running at half peak thruput (it took careful programming to keep 195 operating at peak thruput). The idea behind the dual i-stream was to have two sets of independent instruction streams ... both operating at half peak thruput ... but combined, capable of keeping the 195 pipeline fully stocked with instructions. misc. past posts mentioning dual i-stream as (one of the) mechanism for compensating/masking increasing memory latencies (as measured in processor cycles) http://www.garlic.com/~lynn/94.html#38 IBM 370/195 http://www.garlic.com/~lynn/99.html#73 The Chronology http://www.garlic.com/~lynn/99.html#97 Power4 = 2 cpu's on die? http://www.garlic.com/~lynn/2000g.html#15 360/370 instruction cycle time http://www.garlic.com/~lynn/2001j.html#27 Pentium 4 SMT Hyperthreading http://www.garlic.com/~lynn/2001n.html#63 Hyper-Threading Technology - Intel information. http://www.garlic.com/~lynn/2002g.html#70 Pipelining in the past http://www.garlic.com/~lynn/2002g.html#76 Pipelining in the past
Re: CPU time differences for the same job
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Tom Schmidt) writes: I've been running VM more off than on since PLC 5 and I'm certain that the behavior that I referenced WAS in VM... at some point. But if you Lynn Wheeler say it isn't there now, I'll believe you (unless/until I can prove you wrong, of course). But I know back in the VM/HPO or (maybe) early VM/XA days it was true that VM put itself into a tiny loop while it waited for work. The loop was in a unique-to-VM PSW key so that the hardware monitor (the speedometer) could tell the difference between work and wait. there were a number of specific environment experiments done in that time frame ... for one reason or another. one of the first was for acp/tpf on 3081. acp/tpf didn't support multiprocessor support ... and there wasn't going to be a non-multiprocessor machine. normally to simulate a privilege instruction (not handled by the micrcode) ... interrupt into the vm kernel, do the simulation, and return to the virtual machine. this overhead will tend to be constant from run to run ... directly part of doing work for the virtual machine. over the years, attempts were made to get this as small as possible and/or have it done directly in the hardware of the machine. the other overhead is the cache/paging scenario ... fault for page not in memory and there is overhead to bring the page into memory. this is analogous to cache miss ... and the program appears to execute slower because of the latency to process the cache miss. this can be variable based on other activity going on in the real machine (analogous reasons for both cache misses and page faults). in the acp/tpf scenario ... if essentially just about the only workload was acp/tpf ... the 2nd 3081 processor would be idle. so there was a hack developed for things like SIOF emulation ... interrupt into the kernel, create an asyncronous task for SIOF emulation, SIGP the other processor and return to acp/tpf virtual machine. Creation of asyncronous task, signaling the other processor, taking the interrupts, plus misc. multiprocessing locking/unlocking drove up total avg overhead by ten to fifteen percent. However, the SIOF and ccw translation offloaded to run asyncronously on the idle 3081 resulted in net thruput benefit for the single acp/tpf scenario. The problem was that the implementation drove up the total avg overhead by ten to fifteen percent for every customer running VM on multiprocessor ... even those where the other processors weren't idle. For pure topic drift ... there is something analogous going on in the current environment with multi-core processors being introduced into the desktop/laptop (personal) computing environment. Eventually, 3081 with the 2nd processor was removed was announced as 3083 (for acp/tpf customers). Since 3081 still had the cross-cache chatter 10 percent cycle slowdown scenario i've described for 370 multiprocessor ... they were able to run the single 3081 (aka 3083) processor nearly 15percent faster. even later still, acp/tpf eventually supported multiprocessor. Active wait was another such experiment ... where a specific hardware configuration and workload gained a couple percent if the system effectively polled for something to do. from long ago and far away To: wheeler DATE: 04/19/85 20:58:47 On 4/10/85 xx presented his latest results to management and others and I thought you might be interested to hear how we stack up against HPO. These are runs of VM/XA SF1 (which is Mig. Aid releases 3 and 4 rolled up into one package now), with about 2K LOC of enhancements to boost the performance. The enhancements include processor-local true runlists and active wait, with a master-only runlist also. They also include a significant rework of the drum paging code and rework of the SSKE code (for non-resident pages only?). And other things which I just forget now. All these things collectively saved a whole lot of execution time. As a result, SF1 now can handle 80% of the number of CMS users that HPO can handle, whereas earlier it was only about 60% as many as HPO. ... snip ... now, the HPO base they are referring to still has the 10-15 percent multiprocessor penalty that had been introduced for the acp/tpf environment. There were also a list of a dozen or so other carefully chosen workload and configuration items to try and weight the comparison in VM/XA SF1 favor (CMS workload truely trivial, trivial paging activity, homogeneous well-behaved CMS workload ... but lots of them). I don't remember the exact VM/XA SF1 processor cycles for active wait trade-off vis-a-vis actually being in wait state (and VM/XA was a totally different implementation from VM/HPO). The active wait was along the same lines as the delayed queue drop fix from the same era. There was a bug in identifying idle activity and dropping idle tasks
Re: CPU time differences for the same job
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Phil Smith III) writes: This entire discussion is interesting to a VMer. VM has always done VTIME and TTIME: VTIME is the CPU time used by the guest, TTIME is the VTIME plus CP overhead. VTIME should be repeatable independent of system load; TTIME is what varies. VMers talk about T/V ratios, i.e., How well are we doing in terms of overhead? (Goal=1.00) TTIME overhead is frequently viewed as CPU that would not be there if ran on the real machine (w/o vm). VTIME should be approx. the cpu time running on (bare) real machine (w/o) (although it has all the vaguries and caveats with respect to cache misses, interrupts, etc). the jokes have been (for the other mainframe systems) that if customers could run their applications with and w/o the underlying (mainframe) operating systems ... there would be much less bloat in those systems. in vm case, there was significant more pressure to have highly optimized implementation (because of the with and without issue). as a result ... there were various scenarios over the years ... where if customers could configure their other mainframe operating systems to rely on the underlying VM for various functions ... things actually ran faster than on the bare machine. one of those was VS1 handshaking ... where it was possible to turn off VS1 virtual paging ... and turn over the responsibility to VM ... it wasn't just that it made VS1 run faster (when running in virtual machine) ... but actually could have VS1 running faster ... than when running w/o VM at all. other recent posts related to this subject: http://www.garlic.com/~lynn/2008c.html#78 CPU time differences for the same job http://www.garlic.com/~lynn/2008c.html#80 Random thoughts http://www.garlic.com/~lynn/2008c.html#81 Random thoughts http://www.garlic.com/~lynn/2008c.html#82 CPU time differences for the same job http://www.garlic.com/~lynn/2008c.html#83 CPU time differences for the same job -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: CPU time differences for the same job
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Ted MacNEIL) writes: Most of the time, I had already told them that there was an I/O bottleneck, tape-drive contention, scheduling, etc. issue. But, they told me to upgrade the processor, anyways. Yes, we usually needed the processor, but the other issues would usually give us a better improvement. starting in the mid-to-late 70s ... i was observing that systems were becoming more i/o bound ... and increasing real storage was more and more being used to compensate for i/o limitations. part of the issue was the work that i was doing on dynamic adaptive resource scheduling ... included some stuff to dynamicly adapt to resource bottlenecks. real storage, paging i/o and processor time was becoming less less frequently primary bottleneck. along the way i had made the comment that relative system disk thruput technology had declined by an order of magnitude over a number of years. at some point this drew the attention of some execs in the disk division ... who asked their performance group to refute the statement. after a couple weeks, the group came back and said that i had somewhat understated the situation. part of the analysis was eventually reworked as disk thruput recommendations for a share presentation. old post with acknowledgement from b874 at share 63: http://www.garlic.com/~lynn/2006o.html#68 DASD Response Time (on antique 3390?) old post with summary from presentation b874 at share 63: http://www.garlic.com/~lynn/2002i.html#18 AS/400 and MVS - clarification please other post with reference to b874 http://www.garlic.com/~lynn/2006f.html#3 using 3390 mod-9s in the 80s there was increasing shift to using real storage to compensate for increasing disk i/o thruput limitations. an example is upswing of relational databases. the original relational/sql effort was done on vm370 platform at research http://www.garlic.com/~lynn/subtopic.html#systemr there was some discussions that went on with the ims people in stl (as well as consulting with them) vis-a-vis the difference between 60s dbms and relational dbms. the stl view-point was that relational implicit index doubled the physical disk space requirement and significantly increased the disk i/os for accesses (to process the index structure) ... compared to the direct record pointers (part of data infrastructure) then in common use. the relational position was that the implicit indexes, abstracted away directly exposing record pointers ... significantly reducing the manual maintenance in the care and feeding of dbms. the big upswing in amount of system real storage during the 80s allowed for relational index structure to be cached ... significantly reducing the i/o penalty. disk storage also became significantly cheaper starting to tip the balance between hardware/thruput costs (for relational) vis-a-vis skills manual costs (for 60s dbms technologies). for other topic drift ... posts about getting to play disk engineer in disk engineering (bldg 14) and product test lab (bldg 15) http://www.garlic.com/~lynn/subtopic.html#disk past posts in this thread: http://www.garlic.com/~lynn/2008c.html#78 CPU time differences for the same job http://www.garlic.com/~lynn/2008c.html#80 Random thoughts http://www.garlic.com/~lynn/2008c.html#81 Random thoughts http://www.garlic.com/~lynn/2008c.html#82 CPU time differences for the same job http://www.garlic.com/~lynn/2008c.html#83 CPU time differences for the same job http://www.garlic.com/~lynn/2008c.html#84 CPU time differences for the same job -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: CPU time differences for the same job
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Wertheim, Marty) writes: The discussion on high speed buffers and internal cache misses is right on target. I've got a set of benchmarks that are plain vanilla COBOL programs - no DB2, or anything. One of them steps through 8 tables of 16 MB each (hopefully not your typical COBOL program). On a 2094-717, that program will run in 60 seconds of CPU time when the CEC is 50% busy, 180 seconds of CPU time when the CEC is 95% busy. Other programs using less memory have more stable CPU times, but even with a program using 7MB, CPU times doubled when the CEC got up to 98%. If anyone wants to send me an email off line, I can send you an Excel showing the numbers I've seen. when caches first appeared ... they weren't large enuf to contain multiple contexts ... always loosing on interrupts which resulted burst of high cache misses changing context. one of the things that I did relatively early for large cache 370s ... was dynamically monitor interrupt rate ... and dynamically transition from running applications enabled for i/o interrupts to running disabled for i/o interrupts (when i/o interrupt rate was higher than some threshhold). this was trade-off between timeliness of handling i/o interrupts (responsiveness) and total thruput. above the i/o interrupt rate threshhold ... it was actually possible to process i/o interrupts faster when running disabled for i/o interrupts. running disabled for i/o interrupts tended to significantly improve application cache hit rate ... which translated into higher mip rate (instructions executed per second) and application executing in less elapsed time. at the same time, i/o interrupt handling tended to be batched ... which tended to improve the cache hit rate of the interrupt handlers ... making them run faster. it was possible to experience both increased aggregate thruput and faster aggregate i/o interrupt handling ... when running disabled for i/o interrupts ... and taking i/o interrupts at specific periods. this was part of what i got to ship in my resource manager http://www.garlic.com/~lynn/subtopic.html#fairshare also in the mid-70s, about the time future system got killed http://www.garlic.com/~lynn/subtopic.html#futuresys i was asked to do a smp project (which never shipped) which involved up to five 370 processor configuration ... and extensive microcode capability. as part of leveraging the microcode capability ... i put the multiprocessor dispatching function into microcode ... basically software managed the dispatch queue ... putting things on in specific order ... the multiprocessor microcode would pull stuff off the dispatch queue ... service it and at interrupt/interval ... move it to different queue (this was something similar seen in later intel 432). I also created a similar microcode managed queue for i/o operations ... something like what was later seen in 370-xa. http://www.garlic.com/~lynn/subtopic.html#bounce For standard 370 (two) multiprocessor cache machines ... i did two level operations for cache affinity ... attempting to keep processors executing on the same processor that they were previously executing on (and therefor being able to reuse information already loaded into cache). Traditional 370 (two-way) multiprocessor cache machines slowed down processor cycle by ten percent to accomodate cross-cache chatter between the two processors ... aka a 2-processor machine started out nominal raw thruput as 1.8times a single processor. System software multiprocessor overhead then would further reduce (application) thruput ... so there was typical rule-of-thumb that 2-processor had 1.3-1.5 times a single processor. I did some slight of hand in the system software for multiprocessor and processor affinity ... and had one two-processor configuration where one processor would chug along about 1.5times mip rate of uniprocessor (because of improved cache hit rate ... despite the machine cycle being ten percent slower) and the other processor abot .9times uniprocessor mip rate. The effective aggregate application thruput was slightly over twice uniprocessr ... because of some slight of hand in software implementation and cache affinity. The change-over to 3081 was supposed to only be a multiprocessor machine ... so there was never any need to mention the uniprocessor/multiprocessor hardware thruput differences. In the early 90s, machines started appearing where there were caches in the 1mbyte-4mbyte range ... this was larger than the real-storage sizes when i first started redoing virtual memory and page replacement algoritms as an undergraduate in the 60s ... recent post http://www.garlic.com/~lynn/2008c.html#65 No Glory for the PDP-15 other references http://www.garlic.com/~lynn/subtopic.html#wsclock Some of the major software vendors, like DBMS offerings ... were doing the
Re: Random thoughts ...
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Tom Marchant) writes: I don't know what you mean when you say the cache line was split across domains. I forget whether a line was 32 bytes, but it always cantained the data from a line of storage. The cache was split into an instruction cache and a data cache, though. There was significant logic needed to deal with that. 801/risc (harvard) architecture allowed for independent instruction and data caches ... but didn't provide for hardware cache consistency. this met that self-modifying instructions were out ... and for store-into data caches (which didn't automatically flush every change to storage) ... there were special flush cache instruction for use by loaders (i.e. program loaders have peculiar characteristic that they can be treating instructions as data ... which then have to get out to storage ... before the changes can possibly show up in the instruction cache). amdahl's machine at the time was somewhat viewed as remarkable being able to provide i-cache and d-cache consistency. a few old posting discussing 5880 http://www.garlic.com/~lynn/2006b.html#38 blast from the past ... macrocode http://www.garlic.com/~lynn/2006e.html#31 MCTS http://www.garlic.com/~lynn/2006j.html#35 Code density and performance? http://www.garlic.com/~lynn/2006u.html#27 Why so little parallelism? http://www.garlic.com/~lynn/2006u.html#34 Assembler question http://www.garlic.com/~lynn/2006v.html#20 Ranking of non-IBM mainframe builders? including some old email from the period (which includes reference to 5880 announcement with separate instruction and data cache): http://www.garlic.com/~lynn/2006b.html#email810318 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Random thoughts ...
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. some cache x-over discussion http://www.garlic.com/~lynn/2008c.html#78 CPU time differences for the same job http://www.garlic.com/~lynn/2008c.html#80 Random thoughts in general cache size and cache implementation can affect effective mip rate. there are some number of benchmark applications on the web that attempts to profile and characterize those details. variability concurrent activity, other application executing as well as possible interrupt rates ... can also affect cache hit/miss ratios ... and therefor effective mip thruput ... which, in turns affects measured cpu use. there is also capture ratio issues ... which can also affect measured cpu use. recent post/thread on the subject: http://www.garlic.com/~lynn/2008.html#42 Inaccurate CPU% reported by RMF and TMON there are a lot of common/similar technology issues with dealing with various kinds of caches (virtual memory, database caches, hardware caches, etc). for other topic drift regarding effective thruput with regard to database caches in a loosely-coupled environment ... we looked at this when we were doing scaleup ... old email http://www.garlic.com/~lynn/lhwemail.html#medusa and old post http://www.garlic.com/~lynn/95.html#13 when we were doing ha/cmp http://www.garlic.com/~lynn/subtopic.html#hacmp a lot of dbms caches have been the equivalent of hardware store-thru caches ... both the log and the main dbms being updated about the same time. some of the dbms implementations did an optimization analogous to store-into caches ... sometimes referred to as fast-commit; aka as soon as the log entry was written, the transaction was treated as committed, ... the associated datatbase records writes would be delayed ... being combined with large number of other record writes for optimization of disk arm motion. Also subsequent changes to the same record ... prior to their writes ... would effectively combine multiple changes into smaller number of writes. The problem was that these fast-commit strategies would be negated when working in a loosely-coupled environment. If a transaction came in on a (loosely-coupled) processor ... different from a processor which had a changed record ... not yet written to disk ... before the related locks could be obtained ... all such changed records would first have to be written to their home dbms disk location. As part of doing scale-up work for ha/cmp distributed lock manager ... also worked out being able to do any direct cache-to-cache copies ... piggy-backed with granting lock (in a store-into, fast commit, dbms environment). some of the processor cache implementations had implemented cache-to-cache copies (on miss) ... say, some of the numa hardware (aka we had participated in SCI meetings and consulted with some of the vendors that did NUMA SCI implementations, included one later bought by your favorite mainframe vendor) ... but they didn't have to worry about dbms failure recovery. This can get really tricky attempting to correctly merge logs and do transaction log redo ... in a loosely-coupled environment with multiple independent logs. at the time, we originally worked it out, most of the dbms implementations were still rather dubious that it could work correctly under all possible failure scenarios. random past posts mentioning sci: http://www.garlic.com/~lynn/96.html#8 Why Do Mainframes Exist ??? http://www.garlic.com/~lynn/96.html#25 SGI O2 and Origin system announcements http://www.garlic.com/~lynn/98.html#40 Comparison Cluster vs SMP? http://www.garlic.com/~lynn/2001b.html#39 John Mashey's greatest hits http://www.garlic.com/~lynn/2001b.html#85 what makes a cpu fast http://www.garlic.com/~lynn/2001f.html#11 Climate, US, Japan supers query http://www.garlic.com/~lynn/2001j.html#12 OT - Internet Explorer V6.0 http://www.garlic.com/~lynn/2001j.html#17 I hate Compaq http://www.garlic.com/~lynn/2001l.html#16 Disappointed http://www.garlic.com/~lynn/2002g.html#10 Soul of a New Machine Computer? http://www.garlic.com/~lynn/2002h.html#78 Q: Is there any interest for vintage Byte Magazines from 1983 http://www.garlic.com/~lynn/2002i.html#83 HONE http://www.garlic.com/~lynn/2002j.html#45 M$ SMP and old time IBM's LCMP http://www.garlic.com/~lynn/2002l.html#52 Itanium2 performance data from SGI http://www.garlic.com/~lynn/2003.html#0 Clustering ( was Re: Interconnect speeds ) http://www.garlic.com/~lynn/2003.html#6 vax6k.openecs.org rebirth http://www.garlic.com/~lynn/2003d.html#57 Another light on the map going out http://www.garlic.com/~lynn/2003p.html#30 Not A Survey Question http://www.garlic.com/~lynn/2004.html#1 Saturation Design Point http://www.garlic.com/~lynn/2004d.html#6 Memory Affinity http://www.garlic.com/~lynn/2004d.html#68 bits, bytes, half-duplex, dual-simplex, etc http://www.garlic.com/~lynn/2005.html#40 clusters vs shared-memory (was: Re: CAS
Re: CPU time differences for the same job
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Tom Schmidt) writes: z/VM waits with a CPU loop (so it doesn't need to come out of a wait state when it is waiting) so it would run just as hot when it was idle as otherwise. (Unless there is special code to account for machine perspiration?) Search the archives at http://bama.ua.edu/archives/ibm-main.html arg, if that is true, ... i must have been gone, way too long. virtual machine hypervisor was very careful about waiting ... both because of 1) supporting virtual/virtual ... i.e. it might actually be running in a virtual machine ... so it really might be stealing cycles from other applications 2) lot of work had been done in the 60s to make it cost efficient ... somewhat motivated by various customers using platform for commerial timesharing service bureaus. there was obvious work to make the system operate as efficiently as possible ... aka dispatching, scheduling, paging, pathlengths, etc ... as well as making the processing accounting as accurate as possible. however, there was additional features helping make the transition to offering 7x24 availability of online environment. this started in the period when systems were normally leased and processors had cpu meters ... and system lease charges were based on value accumulated by the cpu meter. one of the tricks developed ... was making sure that the cpu-meter stopped ... when the system was up and available but otherwise idle. other work was enhancing offshift operation when useage might be light or non-existant ... allowing operations w/o onsite operator; aka leave the system up available for offshift, remote logins ... but otherwise minimize as close to zero as possible, cost of system operation. it wasn't just necessary to put the machine into wait state to stop the cpu meter ... but also quiesce all i/o ... but leave the system available for accepting things like incoming keystrokes. One of the idiosyncrasies of the cpu meter operation was that if it was running and everything stopped ... the meter would continue to run for 400 milliseconds before it actually stopped (i.e. for the cpu meter to actually stop, idle had to be for periods longer than 400 milliseconds). trivia question ... what was the wakeup interval for the mvs srm? -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: What happened to resumable instructions?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Paul Gilmartin) writes: o Page faults? How are page faults handled for resumable instructions? Is a fault generated for any page in the range of either operand, with the OS attempting to stage both, or does a fault possibly for each resumption. the resumable instructions were storage-to-storage operations with address and length specified in registers for both source and destination. requirement was that the registers be appropriately updated as the instruction proceeded ... so that on any interrupt, all registers will have been appropriately updated (and therefor on restart, the instruction would resume from the appropriate position). the problem wouldn't so much the psw address being correct on the interrupt ... the issue was that all registers actually reflecting the correct values (i.e. working copies of the values weren't squirreled away in other hardware location). prior to mvcl/clcl ... instructions would pretest start and end location for both (2k) storage protection and 4k page fault. with mvcl/clcl, the testing had to be done incrementally for every byte processed (although there were later optimizations that would do it in larger blocks, and then fall back to per byte mode ... for any residual). highly pipelined machines gets into lots of issues with what are the current (visable) register contents at any point in time (lots of different parallel executing circuits with possibly their own register values). newer machines also have extensive hardware ras with status checkpoint and instruction retry (to mask various kinds of transient hardware glitches) ... instructions that execute incrementally aggrevate status checkpointing overhead (and instruction retry logic). i actually ran into early problem with 370/125 implementation on vm370. at vm370 boot (on real machine), it would load maximum values into mvcl (initialized to clear storage) and kick it off, it would zero all of real storage ... interrupting when it hit the end of real storage. early 370/125 microcode had a bug where it was still pretesting origin end for mvcl/clcl and aborting the instruction before starting ... which to vm370 made it appear like there was no real storage. vm370 was originally targeted at supporting 256kbyte machines ... prior to announce of 370/125 ... and was never announced for 370/125. at this point, a customer had requested assistance in getting vm370 running on 370/125. while vm370 had non-swappable kernel support ... the amount of fixed kernel had somewhat bloated between the original announce ... and this point ... which also significantly aggrevated vm370 operation in 256k real storage. recent post referring to having done lots of work in 60s as undergraduate on cp67 ... much of which got picked up and shipped in the product: http://www.garlic.com/~lynn/2008c.html#65 No Glory for the PDP-15 one of the other things that i had done for cp67 in the 60s was making portions of the kernel pageable ... which wasn't shipped in cp67 product ... but was picked up for vm370 product. however, since the initial release ... things had gotten lax between what was in the fixed kernel and what was in the pageable kernel. by the time of the customer vm370 370/125 request to the science center http://www.garlic.com/~lynn/subtopic.html#545tech I had moved much of the cp67 work (that had been dropped in the cp67 to vm370 morph) as well as adding lots of new stuff ... and was supplying highly modified vm370 systems to large number of internal datacenters. some old email references: http://www.garlic.com/~lynn/2006v.html#email731212 http://www.garlic.com/~lynn/2006w.html#email750102 and part of getting vm370 up and running on 370/125 also involved going thru lots of kernel infrastructure and significantly reducing the fixed storage footprint for running in 256kbyte real storage machine. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Job ad for z/OS systems programmer trainee
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Binyamin Dissen) writes: Well, back when I was starting SP'ing (Western Electric), slightly earlier than your time frame, that was the job. You first figured out the bug, and then reported it to the vendor. There wasn't as much hand holding as nowadays. You were expected to do most of the research yourself. Of course, nowadays with OCO and no fiche, this would be a bit harder. some 40yrs ago, back when i was an undergraduate and doing a lot of virtual machine kernel hacking ... i would get requests for new features from the vendor. in later years, i conjectured (that because of the nature of some of the requests) ... some may have originated from some gov. agency. for some slight topic drift: http://www.nsa.gov/selinux/list-archive/0409/8362.cfm -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Data Erasure Products
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Anne Lynn Wheeler) writes: the issue normally reduces to what is the threat model? security classification tends to be associated with threat model where divulging the information is not desirable ... and classification level attempts to make the measures to prevent information divulging proportional to the damange that might happen if the information is divulged (and/or the effort that an attacker will go to in order to get the data). For magnetic media this might be something like overwritting a specific number of times with (different) random data ... nist standard: http://csrc.nist.gov/publications/nistpubs/800-88/NISTSP800-88_rev1.pdf re: http://www.garlic.com/~lynn/2008c.html#47 Data Erasure Products oh, and note recent article: 'Erased' personal data on agency tapes can be retrieved, company says http://www.govexec.com/dailyfed/0108/012308j2.htm from above: Personal and sensitive government data -- including employees' personal data -- on magnetic tapes that federal agencies erase and later sell can be retrieved using simple technology, according to an investigation conducted by a storage tape manufacturer. ... snip ... the above article references a GAO report/study: According to its September 2007 report (GAO-07-1233R), GAO concluded it could not find any comprehensible data on any of the tapes using standard commercially available equipment and data recovery techniques, specialized diagnostic equipment, custom programming or forensic analysis. ... snip ... i.e. gao report http://www.gao.gov/new.items/d071233r.pdf old article from last sept: Government sale of used magnetic tape storage not a big security risk, GAO reports http://www.networkworld.com/community/node/19807 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Migration from Mainframe to othre platforms - the othe bell?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Steve Comstock) writes: WAIT! STOP! Is an AIX machine a mainframe??? I don't think so. I know the definition is a slippery one, but to me a mainframe is a Syztem z machine, or one of its predecessors (or a competitive machine from the past; maybe Unisys today). there have been a couple different AIXs. there is the risc flavor. research and opd were working on a 801/risc (romp) followon to the displaywriter. when that got killed, the group looked around for something else to apply the machine for ... and settled on the unix workstation market. they got the group that had done the port of att unix for pc/ix to do a similar port. this morphed displaywriter followon was released as pc/rt and the system as aix v2. it becamse aix v3 for rios/power chips on the rs/6000. http://www.garlic.com/~lynn/subtopic.html#801 independent of that there was the port of UCLA's locus unix look-alike to 370. it was sort of a unix SAA strategy ... with the same system announced as aix/370 and aix/386. locus provided for transparent networked filesystem as well as fairly transparent process migration i.e. executing code could move to different processors (even to different kinds of processors ... for some value of transparent). aix/370 was upgraded to aix/esa wiki page: http://en.wikipedia.org/wiki/IBM_AIX_(operating_system) the above is slightly garbled. it mentions much of aix v2 kernel was written in PL/I programming language. pl.8 language and cp.r operating system had been developed for 801/risc and was being used for the displaywriter followon. when that project was killed ... and the machine retargeted for unix workstation ... sort of needed something to do for all the pl.8 programmers. A VRM (virtual resource manager) was defined implemented in pl.8 ... with an abstract virtual machine interface. The vendor that had done the pc/ix port then was instructed to port to the defined abstract interface. The claim was that this could be done faster and with less total resources than having the vendor implement directly to bare metal. This was shown to be incorrect ... and had other implications ... since things like new device drivers ... required both a VRM implementation as well as a AIX implementation. The VRM was dropped as part of AIX V3 and move to RIOS/POWER chips. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Migration from Mainframe to othre platforms - the othe bell?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Scott Lurndal) writes: Locus TNC (Transparent Network Computing). Morphed into OSF1/AD, IIRC. Unisys did a similar OS (SVR4/MK (mk for microkernel, based on Chorus)) for the OPUS product 1989-1997. Became part of the Amadeus project with USL and the EC (can't remember the name of the EC initiative). re: http://www.garlic.com/~lynn/2008c.html#50 Migration from Mainframe to othre platforms - the othe bell? the palo alto group had been working with UCLA on locus and had it installed on series/1 and some 68k machines. they were also working on bsd port for 370. their bsd/370 product was redirected to pc/rt and offered as AOS on the (bare metal) PC/RT (counter example as to the minimal effort it took to do a unix port to the native hardware as opposed to the difficulty involved in dealing with the VRM as in the AIXV2 offering). the palo alto group then produced the aix/370 and aix/ps2 offerings. the corporation equally funded mit project athena with dec ... and each had an assistant director at the project ... which turned out things like kerberos and X (windows, as well as some number of other things) ... recent kerberos reference/post: http://www.garlic.com/~lynn/2008c.html#31 Kerberized authorization service the corporation also funded cmu andrew projects (to the tune equal to the combined corporations' funding at mit) ... which turned out things like andrew filesystem, andrew widgets, mach, and camelot. OSF design meetings were something of a mashup of locus, project athena, andrew, hp/apollo domain, and (at least) aix distributed filesystem. wiki page http://en.wikipedia.org/wiki/Open_Software_Foundation from above: OSF's standard Unix implementation was known as OSF/1 and was first released in 1992.[2] For the most part, it was a failure; by the time OSF stopped development of OSF/1 in 1994, the only vendor using OSF/1 was Digital, which rebranded it Digital UNIX (later known as Tru64 UNIX after Digital's acquisition by Compaq). ... snip ... during this period ... we were sometimes around the cambridge area, but mostly busy with our ha/cmp product (for aix): http://www.garlic.com/~lynn/subtopic.html#hacmp and this old reference to parallel oracle meeting http://www.garlic.com/~lynn/95.html#13 http://www.garlic.com/~lynn/96.html#15 and these old emails on ha/cmp scaleup http://www.garlic.com/~lynn/lhwemail.html#medusa cmu mach (microkernel) showed up in a number of implementations, including NeXT and the current apple operating system ... recent post: http://www.garlic.com/~lynn/2008b.html#20 folklore indeed and wiki page: http://en.wikipedia.org/wiki/Mach_kernel above has reference to A comparison of Mach, Amoeba and Chorus http://www.cdk3.net/oss/Ed2/Comparison.pdf camelot (as encina) was included in the transarc spinoff (along with andrew filesystem) and then directly purchased (joke in previous reference about funding it three different times) ... which has shown up as an aix transaction cics. wiki page ... http://en.wikipedia.org/wiki/Transarc osf was supposedly alternative to att/sun official unix. camelot/encina was an alternative to att tuxedo transaction processing (which was spun off and eventually showed up at BEA, recently purchased by oracel). wiki page http://en.wikipedia.org/wiki/Tuxedo_(software) For a totally different mainframe amadeus drift ... in the late 80s, my wife had done a short stint as chief architect of (this) amadeus (which was being done sort of as a european alternative to the united and american mainframe systems) that started off with the old eastern airline res system http://www.amadeus.com/ and wiki page http://en.wikipedia.org/wiki/Amadeus_CRS other mainframe unixes in the 80s was amdahl uts system and the special product done for internal ATT use, which layered unix interface on top of a stripped down tss/370 kernel. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Migration from Mainframe to othre platforms - the othe bel?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Henry Willard) writes: Other than having AIX in the name and being ports based on some version of Unix, AIX/ESA and AIX/370 didn't have much in common with the AIX that runs on System p. re: http://www.garilc.com/~lynn/2008c.html#50 Migration from Mainframe to othre platforms - the othe bell? http://www.garilc.com/~lynn/2008c.html#53 Migration from Mainframe to othre platforms - the othe bell? one of the issues about the various mainframe unix ports running under vm370 ... was that the effort to add mainframe ras erep to unix base was significantly larger than doing the straight-forward port. running the unix port in a virtual machine ... could leave the ras and erep processing to the base virtual machine kernel. another small issue was getting field engineering sign-off on machine/hardware service unless there was adequate/approved ras and erep. had this discussion about the vast difference in mainframe ras erep and various other systems, with other attendees at this old nasa dependable computing conference: http://www.hdcc.cs.cmu.edu/may01/index.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Young mainframers' group gains momentum
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. Young mainframers' group gains momentum http://www.networkworld.com/news/2008/013108-young-mainframers-group-gains.html Young mainframers' group gains momentum http://www.computerworld.com/action/article.do?command=viewArticleBasicarticleId=9060499 from above: ZNextGen, an organization aimed at young mainframe programmers, has gained significant momentum since it was created roughly two years ago through IBM and its user group, Share, according to its leaders. ... snip ... SHARE zNextGen http://www.znextgen.org/ from above: zNextGen, a user-driven community for new and emerging System z professionals that has the resources to help expediate your professional development skills. ... snip ... -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Data Erasure Products
[EMAIL PROTECTED] (Stephen Mednick) writes: it's not a case of how valuable the data is, more importantly it's to do with what the security classification is that has been assigned to the data. Depending on the data's security classification dictates the media overwriting/sanitisation method that is it be deployed in accordance with government requirements. security classification is simplification ... like role-based access qcontrol is simplification for permissions. ... recent post on dealing with permissions http://www.garlic.com/~lynn/2008b.html#26 folklore indeed the issue normally reduces to what is the threat model? security classification tends to be associated with threat model where divulging the information is not desirable ... and classification level attempts to make the measures to prevent information divulging proportional to the damange that might happen if the information is divulged (and/or the effort that an attacker will go to in order to get the data). For magnetic media this might be something like overwritting a specific number of times with (different) random data ... nist standard: http://csrc.nist.gov/publications/nistpubs/800-88/NISTSP800-88_rev1.pdf An example of how this gets simplified is example of consumer financial information stored at a merchant. The damage gets translated into security proportional to risk ... and the risk is what is the value of the information to the merchant ... old post on the subject: http://www.garlic.com/~lynn/2001h.html#61 Security Proportional To Risk The problem is that the real threat model and therefor risk, is that the value of the information to the consumer (and to any attacking crook) is possibly one hundred times larger than the value of the information to the merchant. The merchant is required to keep transaction logs (and the associated account numbers) for some period as part of mandated business processes. The information value (to the merchant) is some part of the merchant's profit margin on the transaction ... for hypothetical example for some number of product transactions, this could be $10,000. The value of the information to the crook, is related to the credit limits associated with the individual accounts. This could conceivable be $10,000,000 (totally unrelated to the value of the information to the merchant, i.e. some portion of the profit on the purchased products). Since the value to the crook can be 100 to 1000 times larger, the attacking crooks can afford to outspend the defending merchants by possibly one hundred times. in the mid-90s, the x9a10 financial standard working group had been given the requirement to preserve the integrity of the financial infrastructure for all retail payments. Part of this was looking in detail at end-to-end vulnerabilities and threat models ... as part of coming up with x9.59 financial standard http://www.garlic.com/~lynn/x959.html#x959 part of the x9.59 financial standard was eliminating the usefulness of the account transaction log information (at merchants) to attacking crooks i.e. it didn't involve trying to prevent attacking crooks from getting at the information ... it just made the information useless to crooks for performing fraudulent financial transactions. A different example was we also got involved in co-authoring the financial industry x9.99 privacy standard. As part of that we had to look at both GLBA and HIPAA (financial transactions can used for medical procedures which may be listed). One of the issues in HIPAA is that there is a real requirement to make some amount of medical procedure information available. At a result, HIPAA allows for information to be available if it can't be associated with an individual (aka deidentified). deidentified Under the HIPAA Privacy Rule, data are deidentified if either (1) an experienced expert determines that the risk that certain information could be used to identify an individual is 'very small' and documents and justifies the determination, or (2) the data do not include any of the following eighteen identifiers ... snip ... As part of working on x9.99, we put together a privacy merged taxonomy and glossary ... see: http://www.garlic.com/~lynn/privacy.htm for other details see notes at http://www.garlic.com/~lynn/index.html#glosnote -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: New Opcodes
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Tom Marchant) writes: There may have been speculation within IBM that Macrocode, and the architecture that enabled it, was to make it easier to develop new features. I can tell you that I was at Amdahl at the time working on the 580. That was definitely a major reason for it. re: http://www.garlic.com/~lynn/2008c.html#29 New Opcodes http://www.garlic.com/~lynn/2008c.html#32 New Opcodes http://www.garlic.com/~lynn/2008c.html#33 New Opcodes http://www.garlic.com/~lynn/2008c.html#35 New Opcodes well, how should i have phrased it? ... i would run into lots people ... including at the monthly SLAC meetings ... and frequently be asked for advice ... there was lots of issues about not divulging confidences ... even confidences for companies i didn't work for. complicating things, i had a nearly complete set of individually serial numbered (candy striped) 811 documents (i.e. architecture documents named for their nov78 date). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: New Opcodes
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Tom Marchant) writes: It also says, 894 instructions (668 implemented entirely in hardware) The latest POO lists about 750 instructions. I know that there are a few not listed in the POO. Still, it sounds like it's a lot over 50. as per past discussions re the architecture red book (i.e. cms script file where command line option would print the full machine architecture or just the POO subset, full machine architecture was distributed in red 3ring binders) and compareswap instruction http://www.garlic.com/~lynn/subtopic.html#smp getting an instruction added could require a lot of justification. so one way of parsing of the reference to 50+ added instructions to improve compiled code efficiency ... could be referring to over 50 of the added instructions were justified for improving compiled code efficiency (w/o saying anything at all about the total number of added instructions and/or what was the justification for any of the other added instructions). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: New Opcodes
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: as an aside ... there was some similar speculation two decades ago about such stuff. there was even some speculation that one of the other clone processor vendors creation of macrocode was to enable them to quickly adapt to such things (be more agile in tracking, implementing, deploying changes). re: http://www.garlic.com/~lynn/2008c.html#29 New Opcdoes http://www.garlic.com/~lynn/2008c.html#32 New Opcdoes actually such speculation dates back three decades to the introduction of cross-memory instructions and dual-address space mode on 3033 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: New Opcodes
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: actually such speculation dates back three decades to the introduction of cross-memory instructions and dual-address space mode on 3033 re: http://www.garlic.com/~lynn/2008c.html#29 New Opcdoes http://www.garlic.com/~lynn/2008c.html#32 New Opcdoes http://www.garlic.com/~lynn/2008c.html#33 New Opcdoes part of the speculation was that the cross-memory/dual-address space instructions used more STOs (segment table origins) simultaneously ... and the 3033 had inherited its TLB (and STO-associative) implementation from 168. The additional concurrent STO use activity was putting pressure on TLB-miss and therefor performance. one the other hand, large 168 3033 installation were facing enormous pressure on amount of application addressable space ... aka pasts posts about pointer passing paradigm from real memory heritage dictated the SVS and subsequent MVS implementation with the kernel appearing in the application address space. The MVS design included moving (non-kernel) subsystems into their own address space ... dictating the common segment implementation (supporting squirreling away data for pointer passing APIs). Larger installations were having to constantly grow the common segment ... with 24bit addressing (16mbyte), kernel taking up 8mbytes ... and the common segment growing from 4mbytes to 5mbytes (and more) ... was only leave 3-4mbytes (or less) for applications (even tho there was a virtual address space per application). the future system distraction had redirected a lot of effort into non-370 activity http://www.garlic.com/~lynn/subtopic.html#futuresys when future system was killed, there was mad rush to get stuff back into the 370 product pipeline. 370-xa was going to take 7-8 yrs (with 31-bit addressing, access registers, program call return, etc). the stop-gap was 3033 ... which was 168 wiring/logic remapped to faster chip technology. The increasing machine capacity was adding more applications, tending to grow the common segment and putting massive pressure on available (virtual) memory for applications. There was speculation that 3033 cross-memory and dual-address space hardware changes was purely to create incompatibilities for the clone processor vendors ... however there was more than enuf other justification, even if the clone vendors hadn't existed at all (intermediate step on the way to access registers) ... aka dual-address space instructions allowed subsystem to reach directly into the calling application's virtual address to direclty access values pointed to by the passed pointers (w/o requiring the common segment hack). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: New Opcodes
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. re: http://www.garlic.com/~lynn/2008c.html#29 New Opcodes justification is justification ... not all have to be there based on the same justification. as an aside ... there was some similar speculation two decades ago about such stuff. there was even some speculation that one of the other clone processor vendors creation of macrocode was to enable them to quickly adapt to such things (be more agile in tracking, implementing, deploying changes). misc. past posts mentioning macrocode. http://www.garlic.com/~lynn/2002p.html#44 Linux paging http://www.garlic.com/~lynn/2002p.html#48 Linux paging http://www.garlic.com/~lynn/2003.html#9 Mainframe System Programmer/Administrator market demand? http://www.garlic.com/~lynn/2003.html#56 Wild hardware idea http://www.garlic.com/~lynn/2005d.html#59 Misuse of word microcode http://www.garlic.com/~lynn/2005d.html#60 Misuse of word microcode http://www.garlic.com/~lynn/2005h.html#24 Description of a new old-fashioned programming language http://www.garlic.com/~lynn/2005p.html#14 Multicores http://www.garlic.com/~lynn/2005p.html#29 Documentation for the New Instructions for the z9 Processor http://www.garlic.com/~lynn/2005u.html#40 POWER6 on zSeries? http://www.garlic.com/~lynn/2005u.html#43 POWER6 on zSeries? http://www.garlic.com/~lynn/2005u.html#48 POWER6 on zSeries? http://www.garlic.com/~lynn/2006b.html#38 blast from the past ... macrocode http://www.garlic.com/~lynn/2006c.html#9 Mainframe Jobs Going Away http://www.garlic.com/~lynn/2006j.html#32 Code density and performance? http://www.garlic.com/~lynn/2006j.html#35 Code density and performance? http://www.garlic.com/~lynn/2006m.html#39 Using different storage key's http://www.garlic.com/~lynn/2006p.html#42 old hypervisor email http://www.garlic.com/~lynn/2006u.html#33 Assembler question http://www.garlic.com/~lynn/2006u.html#34 Assembler question http://www.garlic.com/~lynn/2006v.html#20 Ranking of non-IBM mainframe builders? http://www.garlic.com/~lynn/2007b.html#1 How many 36-bit Unix ports in the old days? http://www.garlic.com/~lynn/2007d.html#3 Has anyone ever used self-modifying microcode? Would it even be useful? http://www.garlic.com/~lynn/2007d.html#9 Has anyone ever used self-modifying microcode? Would it even be useful? http://www.garlic.com/~lynn/2007j.html#84 VLIW pre-history http://www.garlic.com/~lynn/2007k.html#74 Non-Standard Mainframe Language? http://www.garlic.com/~lynn/2007n.html#96 some questions about System z PR/SM -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Job ad for z/OS systems programmer trainee
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Ted MacNEIL) writes: My degree is a major in computer science with a minor in statistics. My first job was as a capacity analyst. My degree was 100% applicable. a lot of capacity planning came out of a lot of performance work at the science center http://www.garlic.com/~lynn/subtopic.html#545tech ... including modeling and workload profiling and the fundamentals for capacity planning. science center had done the port of apl\360 for cms\apl ... and rather than the toy 16-32kbyte workspaces ... they could be as large as virtual memory/machine size. cms\apl became the basis for much of the sales/marketing support applications on the world-wide hone system (sometime in the early 70s, branch office couldn't even submit mainframe orders that hadn't first been processed by hone application) http://www.garlic.com/~lynn/subtopic.html#hone one of the applications deployed on HONE was the performance predictor (system analytically model from the science center implemented in apl ... allowing branch people to characterize the customer's configuration and workload and then ask what-if questions regarding changes to configuration and/or workload. the production science center online system (first cp67 and then vm370) was heavily instrumented and eventually had 7x24 activity data for approaching two decades and established similar standard for other internal systems. besides the system analytical modeling (including the work that resulted in the performance predictor application) there was also a number of event-driven model implementations. there were also a number of execution sampling implementations ... one which resulted in VS/Repack product in the mid-70s ... which would take trace of instruction address storage references and do semi-automated program reorganization for optimal paging operation. Before release as a product, it was used extensively internally by several products making transition from real-storage environment to virtual storage environment (i.e. for instance, IMS made extensive use of the application). another trace/sampling implementation was also used to determine what functions went into VM ECPS (i.e. 6k bytes of vm370 kernel instructions that represented approx. 70percent of kernel pathlength execution was moved to microcode). http://www.garlic.com/~lynn/94.html#21 another performance optimization methodology used at the science center was multiple regression analysis ... with the detailed 7x24 system monitoring ... it was possible to determine where the system was spending large percentage of its time. i've mentioned before using this methodology on large 450k line cobol application that had been heavily studied and optimized over period of a couple of decades (using products like strobe). it ran on 40+ max configured mainframe CECs ($1.2b-$1.5b aggregate). i used multiple regression analysis to identify another 14percent performance improvement (that hadn't been turned up using the other methodologies). misc. recent references: http://www.garlic.com/~lynn/2006o.html#23 Strobe equivalents http://www.garlic.com/~lynn/2006s.html#24 Curiousity: CPU % for COBOL program http://www.garlic.com/~lynn/2006u.html#50 Where can you get a Minor in Mainframe? http://www.garlic.com/~lynn/2007l.html#20 John W. Backus, 82, Fortran developer, dies http://www.garlic.com/~lynn/2007u.html#21 Distributed Computing -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Peter Relson) writes: You can be certain that POST will always support the CS quick-post protocol and the LOCAL LOCK. re: http://www.garlic.com/~lynn/2008b.html#50 How does ATTACH pass address of ECB to child? as stated in the above post ... the principles of operation wording is from over 35yrs ago ... charlie had invented the compareswap instruction at the science center while doing fine-grain lock for cp67 ... and the pok favorite son operating system people had rejected it ... i.e. the testset was more than adequate for the global kernel spin-lock multiprocessor support from 360 ... mention of global spin-lock http://www.garlic.com/~lynn/2008b.html#58 How does ATTACH pass address of ECB to child? the challenge given the science center http://www.garlic.com/~lynn/subtopic.html#545tech was come up with example uses for compareswap instruction ... other than (multiprocessor) locking operation ... some discussion http://www.garlic.com/~lynn/2008b.html#58 How does ATTACH pass address of ECB to child? i.e. the compareswap instruction can be used for atomic updates for whole class of operations w/o requiring locking operation (which would typically also require supervisor call into the kernel to have uninterruptable operation ... as a means of simulating atomic update). as mentioned in the above analogy with dbms financial transaction and optimistic operation ... that avoids lock ... if certain original conditions continue to hold when the transaction changes are actually made permanent. so the wording in the principles of operation was from the stand-point that the pok favorite son operating system people still had to understand the implications of how compareswap instruction could be used ... and then actually have the changes permeate their implementation. my familiarity with cp67 dates back to when 3 people from the science center http://www.garlic.com/~lynn/subtopic.html#545tech came out to install cp67 at the univ the last week of jan68 (40yrs ago this week) one might claim that maybe that its time to update that 35+yr old perspective in the principles of operation compareswap writeup. A.6 Multiprogramming and Multiprocessing Examples http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.6?SHELF=DZ9ZBK03DT=20040504121320 other posts in this thread: http://www.garlic.com/~lynn/2008b.html#31 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#47 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#48 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#51 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#60 How does ATTACH pass address of ECB to child? for some topic drift ... related to dbms operation ... misc. posts about the original relational/sql project http://www.garlic.com/~lynn/subtopic.html#systemr and transcriptions of '95 SQL reunion http://www.mcjones.org/System_R/SQL_Reunion_95/index.html there is mention of compareswap in the discussion about use for locking http://www.mcjones.org/System_R/SQL_Reunion_95/sqlr95-Shoot-ou.html#Index311 in this session: http://www.mcjones.org/System_R/SQL_Reunion_95/sqlr95-Shoot-ou.html however, the speculation about the origin of compareswap instruction was wrong (i.e. compareswap was chosen because CAS are charlie's initials). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (David Logan) writes: Sorry for the slightly off-topic question, but how come this persons posts are always so complicated to work through with all of the pieces and URL links? Is it the function of how they are posting (i.e. perhaps online with certain options), or is this a manual effort? I'm hoping to not spark a long debate. I'm hoping it's just a one or the other answer. re: http://www.garlic.com/~lynn/2008b.html#63 How does ATTACH pass address of ECB to child? some of it overlaps http://www.garlic.com/~lynn/2008b.html#27 Re-hosting IMB-MAIN and some of it explained in these recent postings http://www.garlic.com/~lynn/2008b.html#57 http://www.garlic.com/~lynn/2008b.html#64 starting in the late 70s, i had been doing semi-automated discussion groups and mailing lists on the internal network http://www.garlic.com/~lynn/subtopic.html#internalnet which was larger than the arpanet/internet from just about the beginning until approx the summer of '85. then somebody packaged up about 300pgs of hardcopy of the discussions, put them in 3ring tandem binders and sent them to everybody on the executive committee (ceo, pres, senior vps). there was also a article in datamation. i got blaimed for all of it. the result was a whole lot of corporate churn and investigations into this new phenonoma. part of results were internal tools deployed to officially support electronic online discussion. major tool could operate both in usenet kind of mode as well in mailing list kind of mode simultaneously (end user could select the option). as noted this was on the internal network and distinct from both bitnet/earn http://www.garlic.com/~lynn/subnetwork.html#bitnet ... although the subsequent listserv on bitnet had some similarities the internal tool. for other topic drift ... old email from the person setting up earn (in europe) looking for computer conferencing tools: http://www.garlic.com/~lynn/2001h.html#email840320 as previously referenced listserv history not started until '86 in europe/earn: http://www.lsoft.com/corporate/history_listserv.asp i have this old joke about in the early 70s, going over to paris to set up a HONE clone (i.e. internal online computing system) ... misc. past posts mentioning HONE http://www.garlic.com/~lynn/subtopic.html#hone as part of the corporate EMEA hdqtrs move from ny to paris ... and having some difficulty reading my email back in the states. one of the other outcomes was that there was a researcher paid to study how i communicate ... they sat in the back of my office for nine months, went to me to meetings, took notes on face-to-face, phone, and electronic communication. they also had softcopy of all my incoming and outgoing email as well as logs of all instant messages. the resulting research was also used as a stanford phd thesis (joint with computer ai and language) as well as subsequent papers and books. misc. past posts related to computer mediated conversation http://www.garlic.com/~lynn/subnetwork.htm#cmc lots of past posts over the last dozen years or so discussing compareswap instruction and/or multiprocessor implementations http://www.garlic.com/~lynn/subtopic.html#smp for slightly related drift as to references, URLs and posting technollgy ... again go back to early days at the science center http://www.garlic.com/~lynn/subtopic.html#545tech not only did the compareswap instruction originate at the science center, but also virtual machine systems, a lot of interactive computing facilities as well as the internal network technology. one of the other things that were invented at the science center was gml (letters chosen for the first letter of the last names of the three people ... although most people are more familiar with generalized markup language) in 1969. gml has subsequently morphed into sgml, html, xml, etc. reference to work at cern morphing sgml into html http://infomesh.net/html/history/early and reference to first webserver outside europe on the slac vm system (cern sister location) http://www.slac.stanford.edu/history/earlyweb/history.shtml for another reference there was this article in ibm systems mag. 3 years ago although some of the details are slightly garbled: http://www.ibmsystemsmag.com/mainframe/marchapril05/stoprun/10020p1.aspx there is also old joke at share circa 1974 ... when cern presented a paper about the results of a cms/tso bakeoff. the company couldn't restrict customer copies of the report ... but internal copies were marked confidential/restricted ... i.e. available on a need-to-know only (wouldn't want to contaminate employees with detailed TSO vis-a-vis CMS comparison). in any case, i've been doing online for almost as long as i've been using cp67 ... 40yrs this week when three people came out the last week jan68 from the science center to install cp67
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: for another reference there was this article in ibm systems mag. 3 years ago although some of the details are slightly garbled: http://www.ibmsystemsmag.com/mainframe/marchapril05/stoprun/10020p1.aspx re: http://www.garlic.com/~lynn/2008b.html#63 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#65 How does ATTACH pass address of ECB to child? i.e. the ibm systems mag article talks about my postings and the archived (URL) references the paper copy of the mag even included a picture of me sitting at home at a keyboard. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Farley, Peter x23353) writes: Granted, but the converse is also true: A posix-style semaphore or queuing mechanism is way overkill for the simplest cases, which IMHO most non-software-house application (not systems) programmers are most likely to need or encounter. Typical in my non-software-house application experience is a single master task with N children, each child with different responsibilities, and each child's only communication need is with the parent, not with any sibling. In this type of application, ECB/POST/WAIT is more than adequate and no requests are ever lost using FastPOST or FastWAIT. Software houses and systems programmers tend to deal with far more complex designs where multiple requestors and/or multiple servers are involved. For those applications a semaphore or queuing mechanisms of some sort are certainly the correct solution. BTDTGTTSTPI. IOW, suit the tool to the job you need to accomplish. ECB/POST/WAIT is a perfectly proportioned mechanism for even quite sophisticated application programming. It's only when you must step up to the next level of multiple requestor/server designs that ECB/POST/WAIT becomes insufficient to your application's needs. all of the large DBMS vendors have gone to some sort of internal task management ... and are using compareswap ... or similar instruction from other hardware vendors (that would offer similar semantics, although the 370 compareswap is the granddaddy). one of the problems that rs/6000 and aix ran into getting all the major vendors to port to the RS/6000 and aix ... was that it failed to have an instruction offering compareswap atomic semantics ... forcing (by comparison) significant performance degradation (using kernel calls). It was possibly initially anticipated that RS/6000 didn't require a compareswap instruction because the rios/power chip didn't offer a multiprocessor option. however, one of the first AIX enhancements for the major DBMS vendors was an emulated compareswap instruction ... which translated into an SVC call with an extremely short instruction emulation fastpath in the supervisor call first level interrupt handler ... with immediate return to application mode. rios/power was always only a single processor but needed to run disabled for all interrupts to provide the emulated atomic compareswap semantics ... and do it with minimal pathlength overhead (thus the special case emulation and return done totally in the svc interrupt handler). there is significant amount of commonality in design across the dbms industry on how to leverage compareswap semantics to implement multithreaded/multitasking operation. lots of past posts mentioning multiprocessor designs and implementations and/or compareswap instruction http://www.garlic.com/~lynn/subtopic.html#smp things get more complex going to cluster (loosely-coupled) operation. my wife had been con'ed into going to POK to be in charge of loosely-coupled architecture. while there, she came up with peer-coupled shared data architecture ... which didn't see a lot of takeup until sysplex (which contributed to her not staying long in the position) ... except for the people doing ims hotstandby ... misc. past posts http://www.garlic.com/~lynn/subtopic.html#shareddata later we did the ha/cmp product http://www.garlic.com/~lynn/subtopic.html#hacmp with a distributed lock manager that managed serialization function transaction semantics across a clustered, loosely-coupled environment. we had extended the work in ha/cmp and distributed lock manager for large-scale scaleup ... mentioned in this old post http://www.garlic.com/~lynn/95.html#13 and some of the scaleup issues discussed in these old emails http://www.garlic.com/~lynn/lhwemail.html#medusa -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Wheeler Postings
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Jon Brock) writes: Here, for instance: http://groups.google.com/group/bit.listserv.ibm-main/msg/7941aee482af5b48? i.e. also archived here http://www.garlic.com/~lynn/2006q.html#25 garlic.com and some related recent references, also about posting http://www.garlic.com/~lynn/2008b.html#65 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#66 How does ATTACH pass address of ECB to child? now, a post on the subject in the original thread: http://www.garlic.com/~lynn/2008b.html#63 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#69 How does ATTACH pass address of ECB to child? mentions work on original relational/sql implementation, including technology transfer to endicott for sql/ds: http://www.garlic.com/~lynn/subtopic.html#systemr also mentioned was scaleup for our cluster/distributed high availability product: http://www.garlic.com/~lynn/subtopic.html#hacmp one of the people in this (ha/cmp) meeting, previously mentioned http://www.garlic.com/~lynn/95.html#13 says that he handled much of the technology transfer from endicott (sql/ds) back to STL for DB2. now two of the other people in that same meeting show up a little later at a small client/server startup responsible for something called the commerce server. we were called in to consult because they wanted to do payment transactions on their server ... and also wanted to use a technology that had been invented called SSL for the implementation: http://www.garlic.com/~lynn/subnetwork.html#gateway that work is now frequently referred to as electronic commerce. 40yrs since science center installed cp67 at the univ. the last week of jan68. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Paul Gilmartin) writes: I have believed, and other updates to this thread appear to concur, that WAIT/POST are older than CS. At some time, then, WAIT/POST code must have used some other locking mechanism. So, after CS first became available there may have been some interval before it was reliable to use CS to bypass POST. When did that interim conclude, making it safe to bypass POST in that fashion? And might there be open-source archival MVS (3.8 or earlier) that could execute on emulated hardware supporting CS, but for which POST requires the actual SVC? re: http://www.garlic.com/~lynn/2008b.html#31 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#47 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#48 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#50 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#51 How does ATTACH pass address of ECB to child? wait/post was svc into the kernel running disabled for interrupts ... and therefor would be serialized from the standpoint of other events. the state of the multiprocessing art when charlie was working on fine-grain multiprocessor locking for cp67 (and invented compareswap instruction) ... was single supervisor/kernel spin-lock ... i.e. a single kernel variable that all first level interrupts handlers would perform a testset on initial entry. if the testset was succesful, execution would continue thru the first level interrupt handler and into the rest of the kernel. if testset was unsuccesful, the code would continue to branch back to the testset instruction until succesful. on leaving the kernel (for application/problem execution), the global kernel spin-lock would be cleared to zeros (allowing other processors to execute in the kernel). implicit was that all wait/post operations were correctly serialized ... along with effectively all other kernel functions ... by combination of the kernel running disabled for interrupts ... and, on real multiprocessor, testset serialization on the global kernel spin-lock. as noted, with other organizations believing that testset was adequate for all multiprocessor functions ... the challenge given the science center http://www.garlic.com/~lynn/subtopic.html#545tech was to come up with compareswap uses that weren't multiprocessor specific. the result was the stuff that appears in the appendex of the principles of opertion ... previous reference http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/dz9zr003/A.6?DT=20040504121320 where it was possible to implement correct, concurrent shared operations w/o requiring the overhead of kernel/supervisor calls (which achieved correctly serialized operation, in part by running disabled for asynchronous interrupts). other past posts mentioning various multiprocessor and/or compareswap interrupt http://www.garlic.com/~lynn/subtopic.html#smp -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (McKown, John) writes: Atomic is used quite a bit in computer science. Like the original Greek, it means indivisible. That is, when an atomic operation occurs, any other process will either see the data in the original form, or in the updated form. But it will never seen an intermediate form. It is especially used in relational databases. This is usually done so that, for instance, if you transfer money from checking to savings, you never see the amount in both places. You either see the amount in checking or in savings. This despite the fact that a lot of stuff is going on to do the move. compare and swap simply means that. It compares field1 against field2. If field1 equals field2, then the contents of field3 is placed in field2. What is atomic is that there is a lock that occurs so that at the instant before field1 is compared against field2, no other process is allowed to modify or even look at field2 until after the instruction finishes. This ensures that once the operation starts that no other process can make any updates or decisions based on the contents of field2. there are ACID properties with regard to transactions ... frequently things like financial. A .. atomicity C .. consistency I .. isolation D .. durability a wiki ref http://en.wikipedia.org/wiki/ACID simpler example is two concurrent transactions, one a debit and one a credit ... attempting to update the *same* value. say you have $5000 in savings account. there is a transaction that fetches the current value ... in order to subtract a $200 ATM debit. At the same time there is another transaction that is attempting to perform (add) an EFT $1000 deposit. W/o the transaction serialization semantics, the EFT $1000 deposit concurrently fetches the current ($5000) value, adds $1000 and starts a store operation of the $6000 value ... concurrently while the ATM debit is performing the store of the $4800 value. correct transaction atomic serialization, should result in correct $5800 when everything is done. w/o transaction atomic serialization, the $6000 store might be done followed by the $4800 store ... or the 4800 store might be done followed by the $6000 store (resulting in either $4800 or $6000, neither correct) for the compareswap scenario. The $4800 store only completes if the value being replaced is $5000 ... otherwise things start all over ... similarly, the $6000 store only completes if the value being replaced is $5000 ... otherwise things starts all over again. in transaction systems, this is sometimes referred to as optimistic ... as opposed to purely serialized locking systems ... which only allow one single transaction at a time to fetch a value ... and no other transaction is allowed to fetch a value (or otherwise proceed) until the active transaction has completed ... this is somewhat analogous to the kernel spin-lock mentioned in previous description: http://www.garlic.com/~lynn/2008b.html#58 How does ATTACH pass address of ECB to child? the transaction properties for consistently updating two different values ... is somewhat more complicated ... than simple compareswap. Doing a transfer from one account to another requires that two values (not just one) be correctly updated as a single, syncronized, serialized operation (w/o allowing other simultaneous transactions to corrupt the values). in a few, carefully controlled situation it can be done by compare-double swap. a savings account value and the debit account value are stored in contiguous storage locations. For a transfer operation, both values are fetched, and the appropriate changes (a subtraction from one and an addition to the other) are made to both values. then compare-double and swap is executed ... only succeeding with the replace of both values IFF (if only if) neither value has changed since the original fetch operation. If either or both values have changed since the initial fetch (because other operations are concurrently attempting to change the same fields) ... then the store operations fail ... and the process starts all over again. the other alternative ... for concurrent updates of multiple different values is to resort to something like the spin-lock scenario ... only allowing one processing to performing operations at a single time. As pointed out ... to avoid getting into trouble ... the active running transaction can't be interrupted while it is performing the locked operation. compareswap semantics tends to provide much higher level of concurrent operations with much lower overhead for providing correct serialized operation. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Paul Gilmartin) writes: This is the Principles of Operation; it is not the Assembler Services manual. Note carefully the provided that clause. It's important. Is there any explicit guarantee in the Assembler Services manual that the supervisor WAIT and POST routines use COMPARE AND SWAP to manipulate [ECBs], and that the ECBs are updated only after all collateral operations are completed? If not, the programmer who engages in such shenanigans is using unsupported interfaces. re: http://www.garlic.com/~lynn/2008b.html#31 How does ATTACH pass address of ECP to child? you have to remember that this was being written over 35yrs ago ... after the favorite son operating system people in pok claimed that the compareswap instruction wasn't necessary (for multiprocessor operation); that testset (carried forward from 360 multiprocessor) was more than satisfactory. http://www.garlic.com/~lynn/subtopic.html#smp the challenge then to the science center http://www.garlic.com/~lynn/subtopic.html#545tech ... in order to get compareswap justification for inclusion in 370 architecture ... was to come up with other uses ... that weren't multiprocessing specific. the result was some number of examples of its use in multiprogramming/multithreaded operation (that wasn't necessarily multiprocessing). in any case, since then there have been several generations, allowing compareswap instruction use to widely permeate through the infrastructure. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (David Logan) writes: Test and set works just fine too. They are both atomic. re: http://www.garlic.com/~lynn/2008b.html#31 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#47 How does ATTACH pass address of ECB to child? the issue with atomic testset was that it was a purely binary value ... locked or not locked. charlie was working on fine-grain cp67 multiprocessing locking at the science center (virtual machine operating system running on 360/67) http://www.garlic.com/~lynn/subtopic.html#545tech when he invented compareswap (as previously noted, CAS was chosen because it is charlie's initials). http://www.garlic.com/~lynn/subtopic.html#smp compareswap ... allowed for full word binary value ... as opposed to simple set/notset (or locked/notlocked) ... and compare-doubleswap allowed for double word value atomic update. in the multithreaded/multiprogramming value ... compareswap allowed for atomic updating of a wide variety of values. the testset scenario allowed for locking a section of code, non-atomic update of value within the locked code section ... and then unlocking the code section. independent executing paths would arrive at the locked code section and spin until the other executable path had released the lock. the problem in the multiprogramming/multithreaded sequence ... is that the executable thread could be interrupted while doing the non-atomic update (but still holding the lock) ... with execution then passing to another thread which went into unending spin-loop (waiting for the suspended thread to release the lock). as noted in the examples added to the principles of operation was examples of atomic updates of more complex values (than simple set/not-set) ... which could be used even in interruptable code. prior to compareswap ... such multiprogramming/multithreaded operation always required the overhead of supervisor call (for performing non-atomic updates in non-interruptable code ... and avoiding possible application spin loops). from long ago and far away A.6 Multiprogramming and Multiprocessing Examples http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.6?SHELF=DZ9ZBK03DT=20040504121320 for other topic drift ... it has now been 40yrs since three people from the science center came out and installed cp67 at the university. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (David Logan) writes: While all of that is interesting, and useful in its own right, none of that matters when a you want to perform a simple atomic test and set of a the posted bit in an ECB. Both instructions are atomic for this purpose. You're not going to have any threading problems when using either instruction to check/set the bit in an ECB. The only reason to stay away from TS when posting is if you also want to update the completion code. And the atomic nature only matters when setting the posted bit. For the task preparing to wait, since it's only a check of the posted bit, you perform a TM and then call WAIT if the posted bit hasn't been set by the time of the TM. the original post http://www.garlic.com/~lynn/2008b.html#31 How does ATTACH pass address of ECB to child was about quote from the principles of operation with regard to the example of post routine bypass: A.6.3.1 Bypass Post Routine http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.6.3.1?SHELF=DZ9ZBK03DT=20040504121320CASE= from above: The following routine allows the SVC POST as used in MVS/ESA to be bypassed whenever the corresponding WAIT has not yet been executed, provided that the supervisor WAIT and POST routines use COMPARE AND SWAP to manipulate event control blocks (ECBs). ... snip ... somebody then noted the caveate about supervisor WAIT POST routines using COMPARE AND SWAP ... and I replied that the original writeup was over 35 yrs ago ... back to when the pok favorite son operating system assumed that testset was more than adequate for all purposes. http://www.garlic.com/~lynn/2008b.html#47 How does ATTACH pass address of ECB to child (atomic) testset http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/7.5.139?SHELF=DZ9ZBK03DT=20040504121320 from above: The leftmost bit (bit position 0) of the byte located at the second-operand address is used to set the condition code, and then the byte is set to all ones. ... snip ... has always been defined to test a single bit and if it is zero ... sets the whole byte to ones (xff) as well as the corresponding condition code. so the post routine bypass needs to do atomic replace of the whole word (or simulated atomic replace by executing in disabled kernel with code function locking). for topic drift, quick search engine turns up this reference to ECB field, wait, post (and other os simulation services): http://www.agorics.com/Library/KeyKos/Gnosis/190.html from above: SVC 2 - POST POST is fully supported. Bit 0 of the ECB is set to 0, bit 1 is set to 1, and bits 8-31 are set to the specified completion code. ... snip ... recent post referencing gnosis and keykos http://www.garlic.com/~lynn/2008b.html#24 folklore indeed but then a little more searching turns up this reference Synchronizing Tasks (WAIT, POST, and EVENTS Macros) http://publib.boulder.ibm.com/infocenter/zos/v1r9/topic/com.ibm.zos.r9.ieaa600/tasks.htm from above: Figure 42. Event Control Block (ECB) 0 1 2 +---+---+---+ | W | P | completion code | +---+---+---+ When an ECB is originally created, bits 0 (wait bit) and 1 (post bit) must be set to zero. If an ECB is reused, bits 0 and 1 must be set to zero before a WAIT, EVENTS ECB= or POST macro can be specified. If, however, the bits are set to zero before the ECB has been posted, any task waiting for that ECB to be posted will remain in the wait state. When a WAIT macro is issued, bit 0 of the associated ECB is set to 1. When a POST macro is issued, bit 1 of the associated ECB is set to 1 and bit 0 is set to 0. For an EVENTS type ECB, POST also puts the completed ECB address in the EVENTS table. ... snip ... i.e. initially both bits zero and one are zero. wait specifies that bit zero is set to one. post specifies that bit 0 is set to zero, bit one is set to one, and the rest of the word is filled in with complettion code. testset checks bit zero for being set to zero and then sets the whole byte to one. ... so discussing the post routine bypass scenario (from the orignal post) ... of a multithreaded operation performing a post operation ... when the WAIT hasn't yet been executed (i.e. both bits zero and one are still zero), an atomic replace occurs setting bit 0 to zero, bit 1 to one and the rest of the word to the completion code (as long both bits zero and one are still zero). this is further dependent on the wait routine only doing an atomic setting of bit zero to one ... as long as both bits zero and one are still zero. unfortunately, testset only tests bit zero for zero ... before setting the whole byte to one. that means that atomic testset doesn't correctly perform the wait operation if the ECB has already been
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (David Logan) writes: While all of that is interesting, and useful in its own right, none of that matters when a you want to perform a simple atomic test and set of a the posted bit in an ECB. Both instructions are atomic for this purpose. You're not going to have any threading problems when using either instruction to check/set the bit in an ECB. The only reason to stay away from TS when posting is if you also want to update the completion code. And the atomic nature only matters when setting the posted bit. For the task preparing to wait, since it's only a check of the posted bit, you perform a TM and then call WAIT if the posted bit hasn't been set by the time of the TM. re: http://www.garlic.com/~lynn/2008b.html#50 How does ATTACH pass address of ECB to child? http://www.garlic.com/~lynn/2008b.html#50 How does ATTACH pass address of ECB to child? aka ... the ECB is defined as testing both bits zero and one ... where atomic testset instruction only tests bit zero. wait semantics defines that it tests that ECB isn't already being waited on (bit zero set to zero) and also tests that the ECB isn't already posted (bit one set to zero). using testset instruction will correctly handle an ECB that isn't already being waited on ... but will incorrectly handle an ECB that has already been posted (since testset instruction *ONLY* tests bit zero before replacing the whole byte with ones). the original operating system convention would always call the supervisor ... so it was non-interruptable when the ECB field was being updated ... and used testset in multiprocessing configuration to serialize execution on the different processors. moving to multiprogramming/multithreaded code w/o code serialization locks and code enabled for interrupts ... required atomic updates of the field ... based on update semantics following the rules for both bits 0 (wait) and one (post) ... not just a single bit ... aka wait processing honors ECB that is already being waited on ... bit 0, and/or has already been posted ... bit 1. testset can only recognize an ECB that is already being waited on ... but won't recognize an ECB that has already been posted ... before it obliterates all the bits in the byte setting them all to one. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: How does ATTACH pass address of ECB to child?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Lindy Mayfield) writes: What doesn't is that Cannatello's book has a page and a half on doing POST, with one example of how to change the ECB without using the POST macro. He even has the child checking the ECB to see if a WAIT had been issued. A.6.3.1 Bypass Post Routine http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.6.3.1?SHELF=DZ9ZBK03DT=20040504121320CASE= from above: The following routine allows the SVC POST as used in MVS/ESA to be bypassed whenever the corresponding WAIT has not yet been executed, provided that the supervisor WAIT and POST routines use COMPARE AND SWAP to manipulate event control blocks (ECBs). ... snip ... i.e. charlie had been working on fine grain multiprocessor locking for cp67 at the science center http://www.garlic.com/~lynn/subtopic.html#545tech when he invented the compareswap instruction http://www.garlic.com/~lynn/subtopic.html#smp note compareswap was chosen because CAS are charlie's initials. trying to get the instruction into the 370 architecture was initially rebuffed since the pok favorite son operating system claimed that testset ... carried forward from 360 multiprocessor days, was all that was necessary. the statement was made that in order to get compareswap into 370 architecture required coming up with uses that weren't multiprocessor specific. came up with the multitasking/multithreaded examples ... which were included in the compareswap programming examples. A.6 Multiprogramming and Multiprocessing Examples http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.6?SHELF=DZ9ZBK03DT=20040504121320 more recently the perform locked operation instruction was defined ... and added to the above description. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Flash memory arrays
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-mai,alt.folkore.computers as well. [EMAIL PROTECTED] (, IBM Mainframe Discussion List) writes: Around mid-1999 EMC signed a multi-$billion contract to buy a HUGE number of little disks from IBM over a period of several years. So IBM was making disks then. And I think that contract expired about half a decade ago. san jose plant site (disk unit) now belongs to hitachi. most recent, hitachi talking about selling off the unit. http://www.garlic.com/~lynn/2007v.html#33 Hitachi, Silver Lake in talks about hard drives, sources say a few other references http://www.garlic.com/~lynn/2003d.html#9 IBM says AMD dead in 5yrs ... -- Microsoft Monopoly vs. IBM http://www.garlic.com/~lynn/2003i.html#25 TGV in the USA? http://www.garlic.com/~lynn/2003n.html#39 DASD history http://www.garlic.com/~lynn/2006.html#21 IBM up for grabs? http://www.garlic.com/~lynn/2006o.html#18 RAMAC 305(?) http://www.garlic.com/~lynn/2006r.html#14 50th Anniversary of invention of disk drives http://www.garlic.com/~lynn/2006r.html#15 50th Anniversary of invention of disk drives http://www.garlic.com/~lynn/2006r.html#20 50th Anniversary of invention of disk drives -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Re-hosting IMB-MAIN
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Thomas Kern) writes: You could save some money by running SLES10 and the linux version of LSoft's LISTSERV product. No z/OS or z/VM necessary. recent posts mentioning listserv history http://www.garlic.com/~lynn/2008.html#75 Rotary phones http://www.garlic.com/~lynn/2008.html#76 Rotary phones on (vm-based) bitnet/earn: http://www.garlic.com/~lynn/subnetwork.html#bitnet including the following reference: 1991 The international BITNET network reached its peak, connecting some 1,400 organizations in 49 countries for the electronic, non-commercial exchange of information in support of research and education. Thanks largely to the volunteer efforts of Eric Thomas, BITNET provided thousands of electronic mailing lists based on LISTSERV. Eric Thomas did not want his software to disappear with the mainframes. Therefore, he started looking for ways to port LISTSERV to other environments, such as VMS and Unix. ... snip ... from this site: http://www.lsoft.com/corporate/history_listserv.asp predating listserv on bitnet was the internal corporate vm-based online conferencing facility that had options that would run in LISTSERV-like mode as well as a USENET-like mode. and predating all of them was the online computer conference that Tymshare provided to share starting aug76 ... archives: http://vm.marist.edu/~vmshare/ on their vm-based commercial timesharing platform http://www.garlic.com/~lynn/subtopic.html#timeshare some old email with vmshare references http://www.garlic.com/~lynn/lhwemail.html#vmshare one of my hobbies was providing custom, highly modified vm systems to internal locations ... including the HONE infrastructure http://www.garlic.com/~lynn/subtopic.html#hone which had sort of started out after the 23jun69 unbundling announcement to provide operating system hands-on experience for people in branch offices (running in virtual machines ... starting out with number of deployed cp67 systems). HONE also evolved some number of cms/apl based sales marketing applications which came to dominant all HONE activity. Eventually HONE clones were deployed all over the world ... and it wasn't even possible to submit a customer order that hadn't first been processed by a HONE application. part of what i was doing with vmshare ... was setting up a process where I replicate all the vmshare files (from tymshare) on the various HONE systems. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Flash memory arrays
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (, IBM Mainframe Discussion List) writes: Around mid-1999 EMC signed a multi-$billion contract to buy a HUGE number of little disks from IBM over a period of several years. So IBM was making disks then. And I think that contract expired about half a decade ago. san jose plant site (disk unit) now belongs to hitachi. most recent, hitachi talking about selling off the unit. http://www.garlic.com/~lynn/2007v.html#33 Hitachi, Silver Lake in talks about hard drives, sources say a few other references http://www.garlic.com/~lynn/2003d.html#9 IBM says AMD dead in 5yrs ... -- Microsoft Monopoly vs. IBM http://www.garlic.com/~lynn/2003i.html#25 TGV in the USA? http://www.garlic.com/~lynn/2003n.html#39 DASD history http://www.garlic.com/~lynn/2006.html#21 IBM up for grabs? http://www.garlic.com/~lynn/2006o.html#18 RAMAC 305(?) http://www.garlic.com/~lynn/2006r.html#14 50th Anniversary of invention of disk drives http://www.garlic.com/~lynn/2006r.html#15 50th Anniversary of invention of disk drives http://www.garlic.com/~lynn/2006r.html#20 50th Anniversary of invention of disk drives -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Flash memory arrays
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Clark Morris) writes: Au contraire. My USB key is FBA formatted in 512 byte sectors (I think it is one of the FAT formats available to Win 98 or earlier). FBA is oriented to both disk and even more so, solid state. There are a number of limitations in CKD that will be painful to eliminate and even if they are we are still left with a KLUDGE for which the phase out should have started 25 years ago. re: http://www.garlic.com/~lynn/2008b.html#15 Flash memory arrays i offered well over 25 yrs ago (i.e. 3370 fba). The response i got back from the data management group was that (at the time) it would still cost $26m for training, education, documentation, etc ... even if i provided fully integrated and tested implementation. the claim was that i wouldn't be able to show the necessary ROI for the $26m since customers would just buy the equivalent in fba that they would have ben spent on ckd (no incremental revenue ... and therefor no ROI). the issues about life-cycle costs with regard to maintaining ckd (and life-cycle savings converting to fba) were discounted. other posts mentioning ckd, fba, multi-track search, etc. http://www.garlic.com/~lynn/subtopic.html#dasd one of the first costs was trying to get some of the eckd kludge to work for various things ... like speed-matching buffer (aka 3880 supporting attaching 3380 3mbyte datastreaming, to 168/3033 1.5mbyte channels). a couple recent posts on the subject: http://www.garlic.com/~lynn/2007e.html#40 FBA rant http://www.garlic.com/~lynn/2007o.html#54 mainframe performance, was Is a RISC chip more expensive? including this old email reference mentioning problems getting eckd for speed-matching buffer working http://www.garlic.com/~lynn/2007e.html#email820907b -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Flash memory arrays
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Clark Morris) writes: There obviously would have to be a co-existence period where both architectures are supported. VSAM is already FBA as are all of the newer data architectures. The challenges will be spool, providing GDG like capability to the VSAM ESDS, moving PDSE read access to the Nucleus and deciding how to provide the current SYS1.NUCLEUS capability. Maybe the MVS people should humble themselves and talk to the VM and VSE people to find out how they solved the problem. re: http://www.garlic.com/~lynn/2008b.html#15 Flash memory arrays http://www.garlic.com/~lynn/2008b.html#16 Flash memory arrays http://www.garlic.com/~lynn/2008b.html#17 Flash memory arrays next week is 40yrs since i started on virtual machines ... i.e. three people had come out from the science center http://www.garlic.com/~lynn/subtopic.html#545tech to install (virtual machine) cp67 at the university (i had already been doing optimization work on os/360 for a couple yrs). From the original implementation in the mid-60s, both cp67 and cms had been logical fba ... even when using ckd dasd (which effectively hasn't changed .. and subsequently made it trivial to support real fba (3310 3370) devices. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Flash memory arrays
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (McKown, John) writes: Only z/OS is stuck on ECKD formatted DASD. z/VM and z/VSE can both run on FBA. z/LINUX can run on FBA and/or on SAN/SCSI DASD. I think that the latest z/VM can also run on SAN/SCSI connected DASD as well. z/OS remains the hold out. For whatever reason that may be. there have enormous problems with ckd (and/or trying to get eckd kludge to compensate for the problems). part of it is configuration support ... i.e. device geometry configuration issues are essentially non-existant in platforms supporting fba ... especially vis-a-vis all the stuff that is periodically seen here just on various 3390 model associated geometry problems. another have been speed-matching ... mentioned in previous post http://www.garlic.com/~lynn/2008b.html#16 Flash memory arrays and/or latency issues. in the same time-frame i originally offered FBA support ... i had also done a channel-extender project for the IMS group in STL. STL was bursting at the seams ... and they needed to move 300 from the IMS group to remote off-site location. The problem was that they deemed that the remote 3270 CMS interactive response was unacceptable compared to what they were getting from local 3270 CMS within the STL bldg. The solution was to get CMS local 3270 terminals (for the IMS group) at the remote site (with local CMS 3270 response) back to the vm370 machines in the stl bldg. This was accomplished with channel extender (from network systems corporation) running over T1 (1.5mbit) link. An unexpected side-effect of this effort ... was not only did the IMS group continue to get local 3270 CMS interactive response ... but the channel extender actually improved overall system thruput and performance. The issue was that these were 168/3033 16 channel systems ... where the 3270 control units and disk controllers were spread out over common channel pool. The problem was that the 3270 control units had extremely high channel busy time for the operations they were performing ... which was interferring with disk thruput activity. The local 3270 control units were moved to remote site and replaced on local channel interface with the channel extended boxes ... which had significantly lower channel busy overhead. The resulting reduced channel busy overhead getting 3270 control units off local channels, improved overall system performance by 10-15percent. Basically, I could pretty trivially support almost any kind of direct channel controller at the remote site ... except for count-key-dasd ... even tho the associated speed-matching mismatch for the channel extender was much larger than the factor of two times that later was being dealt with trying to attach 3mbyte 3380s on 370 1.5mbyte channels ... aka channel-extender local 1mbyte devices running over 1.5mbit T1 connection ... nearly a factor of ten speed-match difference compared to the factor of two speed-match difference for 3880 speed-match implementation. ... again, past posts mentioning fba, ckd, etc http://www.garlic.com/~lynn/subtopic.html#dasd getting local 3270 cms terminal thruput for the IMS group was one of the early efforts in the hsdt effort http://www.garlic.com/~lynn/subnetwork.html#hsdt -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Flash memory arrays
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Ed Gould) writes: You have hit it the head. We had a unit that emulated a 2305 (don't remember which model) it worked well . except when we had a power failure. Then at power up *EVERYTHING* was gone. We had to analyze it and put a vtoc back and then redefine PLPA and then IPL again to get the system to use it. I think my hair started to go gray because of the blasted machine. It was gone within a month. Oh, yes we had an application that *NEEDED* (well at least they thought they did) the 2305. In the month that it was going out, we got the applications people to use VIO. They were extremely happy and that was the end of the beast. If the machine survives through a power blink then I would reconsider it but it would take a lot to do so. there were STC solid state ... and for internal datacenters there was something referred to as a 1655 (several hundred) from a vendor that was using memory chips that had failed normal acceptance tests ... but could still be used in this manner. they were most commingly used as paging devices and so not surviving power failure wasn't an issue. they were surplanted by 3090 expanded storage (and later really large real memory) and disk controller electronic caches (initially 3880-11 and 3880-13). old email discussing 2305, 1655, and stc electronic disk comparison: http://www.garlic.com/~lynn/2007e.html#email820805 old posts mentioning 1655, 3880-11, and/or 3880-13: http://www.garlic.com/~lynn/2000d.html#13 4341 was Is a VAX a mainframe? http://www.garlic.com/~lynn/2001.html#18 Disk caching and file systems. Disk history...people forget http://www.garlic.com/~lynn/2001c.html#17 database (or b-tree) page sizes http://www.garlic.com/~lynn/2001d.html#68 I/O contention http://www.garlic.com/~lynn/2001l.html#53 mainframe question http://www.garlic.com/~lynn/2001l.html#54 mainframe question http://www.garlic.com/~lynn/2001l.html#63 MVS History (all parts) http://www.garlic.com/~lynn/2002.html#31 index searching http://www.garlic.com/~lynn/2002d.html#55 Storage Virtualization http://www.garlic.com/~lynn/2002i.html#17 AS/400 and MVS - clarification please http://www.garlic.com/~lynn/2002l.html#40 Do any architectures use instruction count instead of timer http://www.garlic.com/~lynn/2002o.html#3 PLX http://www.garlic.com/~lynn/2002o.html#52 ``Detrimental'' Disk Allocation http://www.garlic.com/~lynn/2003b.html#7 Disk drives as commodities. Was Re: Yamhill http://www.garlic.com/~lynn/2003b.html#15 Disk drives as commodities. Was Re: Yamhill http://www.garlic.com/~lynn/2003b.html#17 Disk drives as commodities. Was Re: Yamhill http://www.garlic.com/~lynn/2003c.html#55 HASP assembly: What the heck is an MVT ABEND 422? http://www.garlic.com/~lynn/2003f.html#5 Alpha performance, why? http://www.garlic.com/~lynn/2003m.html#39 S/360 undocumented instructions? http://www.garlic.com/~lynn/2004d.html#73 DASD Architecture of the future http://www.garlic.com/~lynn/2004e.html#3 Expanded Storage http://www.garlic.com/~lynn/2004g.html#13 Infiniband - practicalities for small clusters http://www.garlic.com/~lynn/2004g.html#17 Infiniband - practicalities for small clusters http://www.garlic.com/~lynn/2004g.html#18 Infiniband - practicalities for small clusters http://www.garlic.com/~lynn/2004g.html#20 Infiniband - practicalities for small clusters http://www.garlic.com/~lynn/2004l.html#29 FW: Looking for Disk Calc program/Exec http://www.garlic.com/~lynn/2005e.html#5 He Who Thought He Knew Something About DASD http://www.garlic.com/~lynn/2005m.html#28 IBM's mini computers--lack thereof http://www.garlic.com/~lynn/2005m.html#30 Massive i/o http://www.garlic.com/~lynn/2005r.html#51 winscape? http://www.garlic.com/~lynn/2005t.html#50 non ECC http://www.garlic.com/~lynn/2006.html#38 Is VIO mandatory? http://www.garlic.com/~lynn/2006c.html#1 Multiple address spaces http://www.garlic.com/~lynn/2006c.html#8 IBM 610 workstation computer http://www.garlic.com/~lynn/2006c.html#46 Hercules 3.04 announcement http://www.garlic.com/~lynn/2006e.html#45 using 3390 mod-9s http://www.garlic.com/~lynn/2006e.html#46 using 3390 mod-9s http://www.garlic.com/~lynn/2006i.html#41 virtual memory http://www.garlic.com/~lynn/2006j.html#11 The Pankian Metaphor http://www.garlic.com/~lynn/2006j.html#14 virtual memory http://www.garlic.com/~lynn/2006k.html#57 virtual memory http://www.garlic.com/~lynn/2006r.html#36 REAL memory column in SDSF http://www.garlic.com/~lynn/2006s.html#30 Why magnetic drums was/are worse than disks ? http://www.garlic.com/~lynn/2006s.html#32 Why magnetic drums was/are worse than disks ? http://www.garlic.com/~lynn/2006v.html#31 MB to Cyl Conversion http://www.garlic.com/~lynn/2006y.html#35 The Future of CPUs: What's After Multi-Core? http://www.garlic.com/~lynn/2007c.html#0 old discussion of disk controller chache
Re: Computer Science Education: Where Are the Software Engineers of Tomorrow?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Michael Stack) writes: This appeared yesterday: http://www.informationweek.com/news/showArticle.jhtml?articleID=205601557 Nearly 70% of middle school teachers lack education and certification in mathematics, let alone computer and business skills, the National Center for Education finds. this and other aspects/posts in similar thread in a.f.c http://www.garlic.com/~lynn/2008.html#44 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#46 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#56 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#57 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#68 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#73 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#87 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#90 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008b.html#1 Computer Science Education: Where Are the Software Engineers of Tomorrow? -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Computer Science Education: Where Are the Software Engineers of Tomorrow?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. re: http://www.garlic.com/~lynn/2008b.html#2 Computer Science Education: Where Are the Software Engineers of Tomorrow? and related new post ... also in a.f.c ... about newly published report announced by nsf today http://www.garlic.com/~lynn/2008b.html#6 Science and Engineering Indicators 2008 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: File Transfer conundrum
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Gary Green) writes: I lost track of who posted the original inquiry, so take this for what it's worth. If the requirement is in the financial industry, could the communications between the two/various systems use S.W.I.F.T. (Society for Worldwide Interbank Financial Telecommunications)? It's been some time since I wrote anything for SWIFT, but it was extremely secure and most financial institutions should be linked in. When I did some work, it was used in the securities market, primarily for payments, foreign exchange, securities, etc... However, there were rumblings that the SWIFT organization was thinking about opening up the network for other financial transactions; which I took to mean data exchange... home page http://www.swift.com/ swift-2 providing internet capability and opening up for b-to-b; we had been brought in to consult with small client/server startup that wanted to do payment transactions on their server; they also had this technology they called SSL they wanted to use ... and result is sometimes now called e-commerce. part of the effort was something called payment gateway (transition between internet and acquiring networks) http://www.garlic.com/~lynn/subnetwork.html#gateway we then were involved in x9a10 financial standard working group (in the mid-90s had been given the requirement to preserve the integrity of the financial infrastructure for all retail payments) that resulted in the x9.59 financial standard http://www.garlic.com/~lynn/x959.html#x959 some years ago we were also asked to provide some input to the swift-2 (what it was called at the time) specification. Connecting to the secure IP network (SIPN) http://www.swift.com/index.cfm?item_id=2304 from above: SWIFTNet messaging services are provided via SWIFT's secure IP network (SIPN), a highly secure and reliable network. Full redundancy, advanced recovery mechanisms and first class operations and customer support services ensure continuous network availability for SWIFTNet services. ... snip .. SWIFTNet Interfaces Qualification http://www.swift.com/index.cfm?item_id=2451 other refs: Securities Markets Infrastructures http://www.swift.com/index.cfm?item_id=2437 Banking Markets infrastructures http://www.swift.com/index.cfm?item_id=57981 from above: Additionally, SWIFT is now complementing its position in the wholesale, high value clearing market by extending its portfolio of SWIFTNet messaging solutions to the low-value payments and ACH market. ... snip ... -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Computer Science Education: Where Are the Software Engineers of Tomorrow?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Patrick O'Keefe) writes: Since the other thread on this topic went off in a seriously OT direction I'll comment on this thread. That original article seemed to imply that the problem was language- based. I've been out of touch with the educational system(s) far too long to have a really know what is currently taught and how it is taught. I have trouble believing that switching from Java to C, C++, LISP, and Ada is going to fix the problem. (Is Ada common in CS curriculum? I notice the authors work at an Ada development shop. They may be a bit biased.) I think their comment that Java encourages a pick a tool that works mentality may be right on, though. Rick Fochtman's question about Radix Partition Tress would make a good test for CS students. I picture a blank stare on the student's faces, but I would love to be shown wrong. re: http://www.garlic.com/~lynn/2008.html#64 Radix Partition Trees there is independent thread in a.f.c regarding the same article, some of the posts: http://www.garlic.com/~lynn/2008.html#44 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#46 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#56 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#57 Computer Science Education: Where Are the Software Engineers of Tomorrow? http://www.garlic.com/~lynn/2008.html#62 competitiveness part of it related to reduced math requirements and part of it related to java (including some mention of java early days). as noted in the radix partition trees thread ... some of luther's work showed up in mainframe instructions. i had been involved in the original relational/sql implementation, system/r http://www.garlic.com/~lynn/subtopic.html#systemr and technology transfer to endicott for sql/ds. for other topic drift ... one of the people in the meeting referenced here ... had mentioned that they had done much of the work for the technology transfer back from endicott to stl for db2 http://www.garlic.com/~lynn/95.html#13 about the same time as the original relational/sql work, I also was involved doing some stuff with a similar, but different kind of dbms implementation (joint project between some people in stl and the los gatos vlsi group). this had some of the similar objectives as the relational/sql activity ... but significantly relaxed the requirements for structured data definition ... and used radix partition trees for its indexing structure (and the person involved in the two mainframe instructions was brought in to consult on some of the work). there was some differences between the old-style '60s DBMS contingent in STL and the relational/sql contingent ... with the '60 DBMS contingent pointing out that relational/sql typically doubled the physical disk requirements (for the table indexes) and also greatly increased the physical disk i/os (for processing the indexes). the relational/sql contingent countered that the use of indexes was part of eliminating the direct record pointer paradigm (that were characteristic of the '60 DBMS) as well as all the associated administrative overhead. during the 80s, things started to tip towards relational/sql ... with disk cost/byte significantly reduced and significant increases in system real storages (allowing index caching, eliminating many of the additional index disk physical i/os) aka change in hardware cost tradeoff versis administrative/skill overhead. for other drift, a totally independent implementation i use for maintaining the rfc index information http://www.garlic.com/~lynn/rfcietff.htm as well as the merged glossary/taxonomy information http://www.garlic.com/~lynn/index.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Radix Partition Trees
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (W. Kevin Kelley) writes: I see that I screwed up and I owe Luther an apology. It should read ...tightest assembly language programs.. There were few problems with Luther's program (other than figuring out how they worked!). re: http://www.garlic.com/~lynn/2008.html#65 Radix Partition Trees a few old posts mentioning luther: http://www.garlic.com/~lynn/98.html#19 S/360 operating systems geneaology http://www.garlic.com/~lynn/98.html#20 Reviving the OS/360 thread (Questions about OS/360) http://www.garlic.com/~lynn/2001.html#2 A new Remember when? period happening right now http://www.garlic.com/~lynn/2001d.html#28 Very CISC Instuctions (Was: why the machine word size ...) http://www.garlic.com/~lynn/2001h.html#73 Most complex instructions http://www.garlic.com/~lynn/2002.html#14 index searching http://www.garlic.com/~lynn/2002d.html#18 Mainframers: Take back the light (spotlight, that is) http://www.garlic.com/~lynn/2002q.html#10 radix sort http://www.garlic.com/~lynn/2003e.html#80 Super-Cheap Supercomputing http://www.garlic.com/~lynn/2003i.html#58 assembler performance superiority: a given http://www.garlic.com/~lynn/2003i.html#83 A Dark Day http://www.garlic.com/~lynn/2004l.html#10 Complex Instructions http://www.garlic.com/~lynn/2005c.html#35 [Lit.] Buffer overruns http://www.garlic.com/~lynn/2005c.html#38 [Lit.] Buffer overruns http://www.garlic.com/~lynn/2005e.html#37 Where should the type information be? http://www.garlic.com/~lynn/2007l.html#57 How would a relational operating system look like? http://www.garlic.com/~lynn/2007o.html#55 mainframe performance, was Is a RISC chip more expensive? http://www.garlic.com/~lynn/2007u.html#18 Folklore references to CP67 at Lincoln Labs http://www.garlic.com/~lynn/2008.html#68 Computer Science Education: Where Are the Software Engineers of Tomorrow? -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Radix Partition Trees
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Rick Fochtman) writes: Has anyone every seen any doc on using radix partition trees? I'm thinking it may have been one of the rainbow books. I vaguely remember data tree structures and I've got a table search problem that might be the perfect application for a tree-structured data repository. The table might have up to 1,000,000 entries, all in storage, and a balanced n-ary tree has GOT to be faster than using a binary search. The nature of the data is such that a plain old-fashioned list, in sorted order, isn't real amenable to a binary search, either. for the fun of it look at: http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.7?SHELF=DZ9ZBK03DT=20040504121320 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Really stupid question about z/OS HTTP server
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (McKown, John) writes: But on the off chance that I'm wrong, I will ask anyway. We use Windows as our desktop OS blech. One nice thing about it is that when we go to a restricted internal IIS web site, we are automagically logged on to the web site via the Active Directory trust mechanism (as I vaguely understand it). Is there any way to extend this so that when a user goes to our z/OS HTTP web server, they can be automagically logged on to their corresponding z/OS RACF id? We do use RACF on z/OS. We don't have any money for this, so a product (unless it is 100% free-as-in-beer and 100% supported) is out of the question. Yes, this is really a whine from the Windows people again about how unfriendly z/OS is. I wonder if they whine about our Linux and Solaris servers as well? can you say kerberos? ... some windows references: http://technet2.microsoft.com/windowsserver/en/library/b748fb3f-dbf0-4b01-9b22-be14a8b4ae101033.mspx http://www.microsoft.com/windowsserver2003/technologies/security/kerberos/default.mspx some ibm references http://www.redbooks.ibm.com/abstracts/sg246540.html?Open http://publib.boulder.ibm.com/infocenter/dzichelp/v2r2/topic/com.ibm.db29.doc.admin/db2z_establishkerberosthruracf.htm http://www-03.ibm.com/servers/eserver/zseries/zos/racf/pdf/share_03_2001_racf_kerberos_windows.pdf http://www-03.ibm.com/servers/eserver/zseries/zos/racf/kmigrate.html and then there is stuff like: IBM CICS RACF Security and Microsoft Windows Server 2003 Security http://technet.microsoft.com/en-us/library/bb463146.aspx kerberos was originally developed a MIT's Project Athena ...and then became internet standard (GSS) ... and has been adopted by quite a few infrastructures for authentication interoperability ... from my rfc index http://www.garlic.com/~lynn/rfcietff.htm select Term (term-RFC#) in the RFCs listed by section, and then select GSS in Acryonym fastpath ... i.e. generic security service (GSS) see also network services , security 5021 4768 4757 4752 4559 4557 4556 4537 4462 4430 4402 4401 4178 4121 4120 3962 3961 3645 3244 3129 2942 2853 2744 2743 2712 2623 2479 2478 2203 2078 2025 1964 1961 1510 1509 1508 1411 ... selecting RFC number brings up the corresponding summary in the lower frame ... i.e. 5021 PS Extended Kerberos Version 5 Key Distribution Center (KDC) Exchanges over TCP, Josefsson S., 2007/08/17 (7pp) (.txt=13431) (Updates 4120) (Refs 4120) (was draft-ietf-krb-wg-tcp-expansion-02.txt) ... and selecting the .txt=nnn filed (in rfc summary) retrieves the actual RFC. misc. past posts mentioning kerberos and/or pk-init http://www.garlic.com/~lynn/subpubkey.html#kerberos -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Really stupid question about z/OS HTTP server
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Chase, John) writes: I *think* you could do that using digital certificates, but I've only read that part of the RACF doc once and have not tried it (yet). re: http://www.garlic.com/~lynn/2008.html#53 Really stupid question about z/OS HTTP server base infrastructure for all of this has been Kerberos. It was originally developed at MIT's project athena ... which as equally funded by DEC and IBM ... and so we got to go by project athena for periodic project revues. originally kerberos was purely password (aka shared-secret) authentication. however, passwords can be evesdropped and reused ... being shared-secret, the same value is used for both originating authentication and validating authentication ... which leads to lots of vulnerabilities and operational problems (including what happens when humans have to deal with scores or hundreds of unique passwords) http://www.garlic.com/~lynn/subintegrity.html#secrets public keys and digital signatures were originally proposed as addressing some of the short-comings of shared-secret infrastructures. first, there is different value for generating authentication information and validating authentication. this can address enormously growing problems with having to manage large number of unique passwords (security 101 typically requires unique passwords for unique security domains as countermeasure to cross-domain attacks ... which is no longer necessary in public key environment). the original draft of pk-init for kerberos ... simply used public keys and digital signatures ... in lieu of passwords for authentication. http://www.garlic.com/~lynn/subpubkey.html#kerberos in purely certificate-less environment http://www.garlic.com/~lynn/subpubkey.html#certless however, a variety of public key operation has evolved with include something called digital certificates ... and digital certificate mode of operation was eventually also added to the kerberos pk-init draft. digital certificates were developed to address the scenario involving first time interaction between complete strangers (aka the letters of credit/introduction from the sailing ship days ... when the relying party had no other means of obtaining information in first time interaction with complete strangers). The purpose of the digital certificates is to carry certified information regarding total strangers that can't be obtained any other way. the issue in all the major institutional authentication scenarios is that digital certificates are redundant and superfluous ... especially in employer/employee scenario ... since it is rarely the case that an employer is rarely dealing with an employee as a total stranger. in a real digital certificate scenario use for (kerberos) authentication, a total stranger ... that is not otherwise known and/or for which there is absolutely no prior information ... is allowed authorized access to the system ... aka nominally the purpose of the digital certificate paradigm is to carry the information about what the person is allowed to do ... and there is no requirement to have any predefined (system) information regarding the individual (and/or what they are allowed or not allowed to do) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Inaccurate CPU% reported by RMF and TMON
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Jason To) writes: We have encountered some weird problem last week and discovered that the total MVS CPU busy percentage reported by both RMF and TMON were inaccurate. RMF and TMON reported MVS CPU percentage does not match with the total CPU% usage by the jobs running in the system at least in one LPAR, the other LPAR seems to be fine. For example the reported total CPU% was 72% at an interval period but only 40% when we add up all the CPU% of jobs, a disparity of 30%. From the WLM activity report, by comparing it with the total APPL% used divided by the total assigned CPs also produced result of 40+%. Hence, the MVS CPU percentage should have been 40+%. Anyone out there have encountered this problem before? Any reported fix to resolve this problem? Btw, we are still at z/OS v1.4, running in the sysplex. can you say capture ratio? some past posts mentioning effect: http://www.garlic.com/~lynn/2005m.html#16 CPU time and system load http://www.garlic.com/~lynn/2006v.html#19 Ranking of non-IBM mainframe builders? http://www.garlic.com/~lynn/2007g.html#82 IBM to the PCM market http://www.garlic.com/~lynn/2007t.html#23 SMF Under VM really strange the first time you here it ... say the elapsed minus measured wait is much larger than the individually measured cpu useages (especially having vm background where it actually does capture nearly every cycle). the referenced SMF Under VM post includes some number of corporate MVS URLs that go into much more detail. old email http://www.garlic.com/~lynn/2006v.html#email800717 discussing moving workload from 168 to 4341s w/o taking into account capture ratio. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: IBM LCS
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Ben Alford) writes: I worked with an IBM 360/65 that had one MB of IBM LCS and later 2 MB of LCS from some other OEM when I was a student. This 360/65 had 3 frames of main storage, each with 256 KB of storage. The way I remember it, we could ask for more LCS than main, but it was much slower. You could specify that data in COMMON as well as data buffers reside in the LCS under OS MVT. The LCS slowdown was noticable because the CPU had to wait longer for all data coming in/out of LCS and, of course, the CPU time was still ticking. I don't know if it slowed down channel access much. i believe had cornell had 360/75 with 8mbytes of ampex lcs. standard 360/75 (and 360/65) storage was 750 nsec for 8byte interleaved access (some claim that 360/75 was a hardwired 360/65 to get extra thruput). some installations setup LCS as extension of (executable) main memory ... that just ran slower. other installations used it for emulated electronic disks ... with emulated data transfers; this could be things like hasp buffer records and/or executable images. i believe the ampex lcs had 8msec access (better than 10 times slower). if you look in 360/65, 360/67, and 360/75 functional characteristic instruction timings ... one of the things will be a prorated part of 750 nsecs for (the 8byte) instruction fetch i.e. 2byte instructions will include 1/4th of 750nsecs for instruction fetch, 4byte instructions will include 1/2th of 750nsecs for instruction fetch, and 6byte instructions will include 3/4th of 750nsecs for instruction fetch. this can be somewhat inaccurate for branch operations ... a double-word aligned (2byte) BR instruction ... will incur the full 750nsecs instruction fetch ... since the rest of the double word won't be used for any other instruction operations. LCS was both an issue of slower electronic memory as well as longer physical signal latency. by comparison 3090 expanded storage was the same electronic memory but physical packaging resulted in longer signal latency. as a result, it was purely packaged as sort of emulated i/o transfer ... but with a much wider bus (so the latency was amortized over much larger amount of data) and syncronized transfer instructions ... to eliminate the significant pathlength overhead in MVS for asynchronous i/o operations. later physical packaging (and improved caching infrastructure) eliminated the need for expanded storage ... however, emulated expanded storage (as part of LPAR configuration) lingered on, compensating for other system issues. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: JCL parms
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Shmuel Metz , Seymour J.) writes: No. The original design of OS/360 was that all programs were subroutines. A program written to be called as a jobstep task could also be called via LINK or ATTACH from another program. It was never designed for programs to assume that they were jobstep tasks, although IBM may have written code[1] that assumed that. some number of univ. developed their own student job monitors (predating availability of watfor). we had 709 running fortran monitor for student jobs ... running tape-to-tape ... tapes were physical moved back and forth between 709 and 1401 ... with the 1401 handling front-end unit record processing. student job elapsed time was on the order of a second. moving to 360/65 with os360 and hasp ... minium elapsed time for 3step (student) fortran compile, link-edit and go ... was on the order of 30seconds ... effectively all of it (constantly re)executing job scheduler. various univ one-step job monitors attempted to attach compile, link-edit, and go for student jobs (eliminating job scheduler overhead). i had done some custom optimization with very careful reorganization of stage-II sysgen output cards ... in order to very carefully physically place files and PDS members on disk ... for optimal arm seek operation. this had improved 3step (effecitvely job scheduler) from 30seconds to about 13seconds (for typical student fortran job). part of old presentation at aug68 SHARE meeting in boston discussing very careful os360 stage-2 sysgen optimization ... in addition to separate activity rewritting large amounts of cp67 virtual machine kernel http://www.garlic.com/~lynn/94.htmL#18 CP/67 OS MFT14 of course when watfor became available ... it eclipsed a lot of the work on one-step job monitors. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: z/OS and VM Control Blocks
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Lindy Mayfield) writes: It was only a question! (-: I certainly didn't mean to upset the status quo. ((--:: recent post from how i tried to handle it long ago and far away ... before OCO in a ipcs alternative that i had implemented in rex(x). http://www.garlic.com/~lynn/2007v.html#46 folklore indeed it was eventually in use by all internal locations and PSRs ... even tho there was a decision not to release it to customers. http://www.garlic.com/~lynn/subtopic.html#dumprx it sort-of started out as a demonstration of the functionality of the new rex ... the stated objective was in half-time over period of 3months, i would re-implement ipcs in rex(x) with ten times the function and it would run ten times faster (little slight of hand since the base ipcs was all implemented in assembler). i had access to softcopy of all the base source files (including control block definitions) and documentation. however, nearly all this stuff had been created for hardcopy/printed output. the particular issue was how to come up with online appropriate information display including being able to tailor to problem being dealt with (very crude online context sensitive orientation). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: For the History buff's an IBM 5150 pc
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Schwarz, Barry A) writes: The Apple I went on sale in l976 so the author seems to have limited view of what a PC is. 5100 pc was announced sep75. http://www-03.ibm.com/ibm/history/exhibits/pc/pc_2.html predating the 5150 pc in 1981. http://www-03.ibm.com/ibm/history/exhibits/pc/pc_1.html from above: One of the earliest IBM attempts to move computing into the hands of single users was the SCAMP project in 1973. This six-month development effort by the company's General Systems Division (GSD) produced a prototype device dubbed Special Computer, APL Machine Portable (SCAMP) that PC Magazine in 1983 called a revolutionary concept and the world's first personal computer. To build the prototype in the short half-year allowed, its creators acquired off-the-shelf materials for major components. SCAMP could be used as a desktop calculator, an interactive APL programming device and as a dispenser of canned applications. The successful demonstration of the prototype in 1973 led to the launch of the IBM 5100 Portable Computer two years later. ... snip ... of course, one could claim that work by science center http://www.garlic.com/~lynn/subtopic.html#545tech creating cp67 virtual machines in the mid-60s, enabled the deployment of CMS personal computing. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Thompson, Steve) writes: WANG with their WANG/VS systems came up with an idea that would have met your problem. The workstations had a button that caused the CEC to download microcode for DP/WP (Data Processing / Word Processing). So your workstation could switch between types of work. [ASCII based S/360 type architecture on steroids.] IBM attempted to make a product to market against WANG. They couldn't figure out how to do it economically. Problem was the distance from the tree to the eyeballs of the powers that were was about 2 inches. The problem was solved by having a PC that could do word processing while having an emulator (3270) for the 3270 type tasks, and then the PC would handle the word processing type tasks. for some topic drift ... past posts about PC getting early market traction because it sold for about the same price as 3270 and in single desktop footprint could get 3270 terminal emulation and some local (personal) computing http://www.garlic.com/~lynn/subnetwork.html#emulation OPD's displaywriter was in WANG wordprocessing market segment. ROMP was early 801 risc chip originally designed to be used for displaywriter follow-on product. when that was killed, the group looked around for something else to use the machine for and settled on the unix workstation market. they got the company that had done the pc/ix port to do one for the displaywriter follow-on and renamed the product the PC/RT (and the software AIX). http://www.garlic.com/~lynn/subtopic.html#801 The PC/RT followon was the RS6000 with RIOS chipset. RS6000 was relogo'ed as hardware platfrom by some number of other companies ... including WANG as it got out of the hardware business. As part of that change-over, some number of the people from RS6000 group went to WANG. old time article from nov80 mentioning wang, word-processing market http://www.time.com/time/magazine/article/0,9171,950498,00.html?iid=chix-sphere page mentioning some of the old/70s wordprocessing market http://www.computermuseum.li/Testpage/DedicatedWPMicros.htm article on demise of dedicated wordprocessor boxes; having given away to multi-application PCs http://www.cbronline.com/article_cg_print.asp?guid=265D4108-6F66-49EC-80B1-E51D2AA8876E note that there was a project in the early 80s to replace the wide variety of internal microprocessors with 801/risc processors (including the ones used for displaywriters). this included all the processors in the low and mid-range 370s ... at the time, the 4341-followon (4381) was going to use a 801/risc processor; the s/38-followon (as/400) was going to use a 801/risc processor ... and lots of others were also. A special flavor of 801/risc, Iliad had additional features for supporting emulation of other architectures ... some old 801-related email, including mention of work on Iliad chips http://www.garlic.com/~lynn/lhwemail.html#801 for other topic drift, old email mentioning 43xx ... e-architecture machines http://www.garlic.com/~lynn/lhwemail.html#43xx i.e. while the high-end 370 came up with 370-xa (code named 811 for nov78 document publication date), the low/mid range came up with e-architecture (where dos/vs to vse came from). for some archeological trivia, i contributed to the document killing 801/risc idea for the 4341-followon. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Ed Gould) writes: Rick: FYI according to [EMAIL PROTECTED] the model 85 was the 1st 360 to have a HSB. re: http:/www.garlic.com/~lynn/2007v.html#98 It keeps getting uglier not just me: http://www-03.ibm.com/ibm/history/history/year_1968.html from above: Additions to System/360 family are announced, including the Model 85. The high-speed cache, or buffer memory, found in the System/360 Model 85, is the first in the industry. The cache memory makes highly prioritized information available at 12 times the speed of regular, main-core memory. ... snip ... not only first 360 ... but first in the industry. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Ed Gould) writes: Its been at least 30 years I will yield to your memory. I just remember it giving me a blow by blow description on the format of the instructions and how it worked plus timings. If it was another manual, fine. 360/30 functional characteristics ... see if this is what you remember http://bitsavers.org/pdf/ibm/360/funcChar/GA24-3231-7_360-30_funcChar.pdf This has directory for -0, -6, and -7 360 Principles of Operation http://bitsavers.org/pdf/ibm/360/poo/ past posts in thread: http://www.garlic.com/~lynn/2007v.html#21 It keeps getting uglier http://www.garlic.com/~lynn/2007v.html#68 It keeps getting uglier for other drift, directory of some 360 FE manuals http://bitsavers.org/pdf/ibm/360/fe/ -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Ed Gould) writes: To refresh my memory was the 370 the first public machine that used the HSB? My memory says yes but as we have seen the POPS and FUNC manual are indeed different. There were quite a few machines I had no exposure to like the 44 and the 67 and the (1)95 among others 360/85 was first machine with cache. this web page has some number of ibm product announcements http://ed-thelen.org/comp-hist/IBM-ProdAnn/index.html including the 360/85 http://ed-thelen.org/comp-hist/IBM-ProdAnn/360-85.pdf -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Doug Fuerst) writes: How do you figure that reverse engineering is an acceptable method of RD or design? Reverse engineering is an easy way to replicate a design. Since the company creating the product, in this case IBM, spent millions developing the machine, they would be entitled to some exclusivity. How fair is it for every competitor to reverse engineer their machines to mimic the IBM box, and not compensate IBM for that? At least MOBO manufacturers use different chipsets and moderately different designs. I don't believe they are reverse engineering Intel boards, nor is AMD reverse engineering Core Duo's. clone controller business was supposedly primary motivation for the future system project ... lots of past posts http://www.garlic.com/~lynn/subtopic.html#futuresys i've posted before about being undergraduate and trying to get the 2702 communication controller to do some stuff and it turned out it couldn't ... which was somewhat motivation for the univ. to start a clone controller project ... reverse engineering the ibm channel interface and building a channel interface card for Interdata/3 ... programmed to emulate 2702. this was written up blaiming four of us for some part of the clone controller business http://www.garlic.com/~lynn/subtopic.html#360pcm article from former corporate executive ... including some number of comments about future system project http://www.ecole.org/Crisis_and_change_1995_1.htm including the following: IBM tried to react by launching a major project called the 'Future System' (FS) in the early 1970's. The idea was to get so far ahead that the competition would never be able to keep up, and to have such a high level of integration that it would be impossible for competitors to follow a compatible niche strategy. However, the project failed because the objectives were too ambitious for the available technology. Many of the ideas that were developed were nevertheless adapted for later generations. Once IBM had acknowledged this failure, it launched its 'box strategy', which called for competitiveness with all the different types of compatible sub-systems. But this proved to be difficult because of IBM's cost structure and its RD spending, and the strategy only resulted in a partial narrowing of the price gap between IBM and its rivals ... snip ... above also referenced here http://www.garlic.com/~lynn/2007u.html#17 T3 Sues IBM To Break its Mainframe Monopoly there was recent question about some number of people departing and going to work on vax/vms ... which led to joke about head of POK having been a major contributor to VMS ... long winded story involving termination of Future System project and mad rush to get stuff back into the 370 product pipeline: http://www.garlic.com/~lynn/2007v.html#96 source for VAX programmers http://www.garlic.com/~lynn/2007v.html#100 source for VAX programmers there is some case to be made that the Future System distraction and leting the 370 product pipeline dry up contributed to giving the processor clones a foothold in the market. past reference to amdahl giving a talk at mit in the early 70s that may be at least partially construed as referring to this ... recent reference http://www.garlic.com/~lynn/2007t.html#68 T3 Sues IBM To Break its Mainframe Monopoly and other parts of postings in that thread: http://www.garlic.com/~lynn/2007t.html#69 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007t.html#71 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007t.html#76 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007t.html#77 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007u.html#1 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007u.html#2 T3 Sues IBM To Break its Mainframe Monopoly -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
virtual appliance
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (David Boyes) writes: Put bluntly, see my other note on using CMSDDR. Getting a 1 pack VM system up on the Flex box works MUCH better, CMSDDR understands CMS file structure so you can just pass images of the volumes over, and as long as you restore the entire volumes, z/OS won't even know it happened. Useful note: when you FTP the CMSDDR output files between the VM system on the Flex box and the new system, use TYPE E, MODE B before you do the PUT or GET in the FTP session. This tells the FTP server on the other end to preserve character set and file parameters, so you won't have to worry about it. Also, keep in mind that your Flex system can create AWSTAPE format files, and that CMS has a AWSTAPE pipe stage that can feed that to CMS utilities. If you're worried about licensing for VM on the Flex box, I think VM/370 (which IBM does not complain about usage) will run CMSDDR. I'll have to look into creating a IPLable system image for people in your situation. Shouldn't be too hard. ... for other topic drift ... in current virtualization genre there is a lot being made of virtual appliance ... somewhat akin to what we use to call service virtual machines but also gaining a lot of cross-over with a system image that can be quickly and trivially deployed ... usually for very specific purposes. In the security genre, such images can also trivially go *poof* (and any compromises, possibly because of an internet connection, evaporates at the same time). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (John P. Baker) writes: The Diagnose instruction has been documented in every Principles of Operation manual issued for the S/360 architecture and for all subsequent superseding architectures, and in every case, has specifically stated that the functions performed by the Diagnose instruction are not published, but may impact any and all aspects of system operation, and if invoked by a user application built without access to that unpublished documentation, may negatively impact the proper functioning of the machine, requiring a Power On Reset and/or the assistance of a Hardware Support Engineer to bring the system back into proper working order. re: http://www.garlic.com/~lynn/2007v.html#21 It keeps getting uglier it also says that the operation of the diagnose may be model dependent. less than a month since it has been 40yrs since I was introduced to (virtual machine) cp67 system ... three people came out to the university from cambridge science center http://www.garlic.com/~lynn/subtopic.html#545tech to install cp67. while an undergraduate I did a lot of rework and optimization of the cp67 kernel. i had also done a lot of work on os/360 optimization ... for the workload at the univ. i gave a presentation at the aug68 share meeting in boston on some of that work ... part of that presentation http://www.garlic.com/~lynn/94.html#18 CP/67 OS MFT14 one of the other things i did was develop a fast-path ccw translation for cms disk i/o when running in a virtual machine (original cms was implemented to be able to run on bare 360/40). I did this by defining a new channel program op-code for disk read/writes ... which acted as an immediate operation ... held the virtual SIO busy until the operation had completed and then presented CC=1, CSW STORED. I got some grief from the people at the science center since i was violating the 360 principles of operation. however, it was a useful performance improvement ... and so it was explained to me that I could use the diagnose instruction ... since the diagnose instruction was defined as being model dependent ... and for CP67 ... and artificial virtual machine 360 *model* could be defined where the diagnose instruction acted as defined by CP67 (w/o violating the principles of operation). misc. past posts mentioning model dependent diagnose instruction: http://www.garlic.com/~lynn/96.html#23 Old IBM's http://www.garlic.com/~lynn/2001b.html#32 z900 and Virtual Machine Theory http://www.garlic.com/~lynn/2002d.html#31 2 questions: diag 68 and calling convention http://www.garlic.com/~lynn/2002h.html#62 history of CMS http://www.garlic.com/~lynn/2003.html#60 MIDAS http://www.garlic.com/~lynn/2003k.html#52 dissassembled code http://www.garlic.com/~lynn/2003m.html#36 S/360 undocumented instructions? http://www.garlic.com/~lynn/2003p.html#9 virtual-machine theory http://www.garlic.com/~lynn/2004.html#8 virtual-machine theory http://www.garlic.com/~lynn/2004d.html#66 System/360 40 years old today http://www.garlic.com/~lynn/2004q.html#72 IUCV in VM/CMS http://www.garlic.com/~lynn/2005b.html#23 360 DIAGNOSE http://www.garlic.com/~lynn/2005j.html#54 Q ALLOC PAGE vs. CP Q ALLOC vs ESAMAP http://www.garlic.com/~lynn/2007f.html#33 Historical curiosity question http://www.garlic.com/~lynn/2007p.html#72 A question for the Wheelers - Diagnose instruction -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: 2007 Year in Review on Mainframes - Interesting
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Conmackie, Mike) writes: And the money I've paid into Social Security all my life will be returned in my retirement with interest ! ss is pay as you go system ... not a fully funded retirement plan. it is one of the reasons why they are concerned about the ratio of people paying-in to the number supported on retirement. SS historical ratio table 1940-2006: http://www.ssa.gov/history/ratios.html This can drastically tip with baby boomers moving from paying to collecting. The first baby boomer collects social security http://abcnews.go.com/WN/LifeStages/story?id=3732745page=1 there is also some gimick on how much is paid, it is currently 15.3% ... but for standard salary workers ... the company has to pay half of it over and above the salary ... and then there is the other half deducted from the salary. This is readily seen in tax returns for self-employed workers where they have to pay the full 15.3%. for most purposes, eliminate the facade and have it restructured so the employers paid the full 15.3% before paying salary (theoritically reducing salaries paid correspondingly) ... with it never showing up for individual employees at all. in past 10-15 there have been some number of companies going under (and/or declared bankruptcy) because their pay as you go retirement systems sometimes reached their largest single expense http://www.skeptically.org/curpol/id7.html ... and federal gov. having to assume the payment http://en.wikipedia.org/wiki/Pension_Benefit_Guaranty_Corporation some number of posts related to unfunded liabilities growing to largest part of the budget and swamping the federal gov ... even if everything else in the budget is eliminated. http://www.garlic.com/~lynn/2007j.html#91 IBM Unionization http://www.garlic.com/~lynn/2007j.html#93 IBM Unionization http://www.garlic.com/~lynn/2007s.html#1 Translation of IBM Basic Assembler to C? http://www.garlic.com/~lynn/2007t.html#13 Newsweek article--baby boomers and computers http://www.garlic.com/~lynn/2007t.html#18 Newsweek article--baby boomers and computers (federal) comptroller general (appointed in the mid-90s for 15yr term) has been making references that congress for at least the past 50 yrs has been capable of simple middleschool arithmatic; recent reference: http://www.garlic.com/~lynn/2007q.html#7 what does xp do when system is copying -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: It keeps getting uglier
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Phil Payne) writes: Has anyone from the Hercules team read IBM's rather stunning admission (on the above page - paragraph 176) that there is a confidential version of the PoP? Their words, not mine. there has been the (confidential) architecture redbook (distributed in red 3ring binders) ... implemented in (cp67/)CMS script file ... with conditional formating to produce either the princples of operation subset ... or the full (confidential) architecture redbook. for other topic drift ... recent post mentioning cms script, gml, sgml, html, system/r, rdbms ... etc http://www.garlic.com/~lynn/2007v.html#17 Amazon's Simple Database misc. past postings mentioning architecture redbook http://www.garlic.com/~lynn/2003f.html#52 ECPS:VM DISPx instructions http://www.garlic.com/~lynn/2004b.html#57 PLO instruction http://www.garlic.com/~lynn/2004c.html#6 If the x86 ISA could be redone http://www.garlic.com/~lynn/2004k.html#45 August 23, 1957 http://www.garlic.com/~lynn/2005b.html#25 360POO http://www.garlic.com/~lynn/2005i.html#40 Friday question: How far back is PLO instruction supported? http://www.garlic.com/~lynn/2005j.html#39 A second look at memory access alignment http://www.garlic.com/~lynn/2005j.html#43 A second look at memory access alignment http://www.garlic.com/~lynn/2005k.html#1 More on garbage http://www.garlic.com/~lynn/2005n.html#48 Good System Architecture Sites? http://www.garlic.com/~lynn/2005p.html#45 HASP/ASP JES/JES2/JES3 http://www.garlic.com/~lynn/2006s.html#53 Is the teaching of non-reentrant HLASM coding practices ever defensible? http://www.garlic.com/~lynn/2007f.html#7 IBM S/360 series operating systems history http://www.garlic.com/~lynn/2007i.html#31 Latest Principles of Operation http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' http://www.garlic.com/~lynn/2007u.html#30 folklore indeed -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: About 1 in 5 IBM employees now in India - so what ?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] writes: Nothing wrong with India - but selfishly, I want jobs where I am, even though I have it better off than those who need jobs there. Of course, in a global economy, you have a lot better chance to sell your wares in countries that you spend money in. there is also issue that knowledge work is pretty distance insensitve in a global economy ... and knowledge work frequently is one of the highest valued work. recent posts about recently published study on educational ranking of different countries http://www.garlic.com/~lynn/2007u.html#78 Educational ranking http://www.garlic.com/~lynn/2007u.html#80 Educational ranking -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: The future of PDSs
[EMAIL PROTECTED] writes: IBM created the PDS a long time ago - giving us some conveniences that fit within its OS design.Other computer companies either did not see this advantage or had OS structures that handled it other ways. Do we use PDSs now because that's what we have been using for decades? Or is it possible to still keep advantages of our OS and go in a different direction? part of the original PDS design was trading off i/o resource vis-a-vis real-storage resource ... at a time when real storage was extremely constrained. some number of other systems ... especially those that came along later when real storage was a much less constrained resource ... made other kinds of implementation tradeoffs -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Shmuel Metz , Seymour J.) writes: The only operating systems that are legal to run on Hercules are Linux, and MVS 3.8 (I think). Shirley all of these are legal: BOS/360 BPS/360 CALL/360 CP/67 DOS/VSE DOS/360 MTS OS/VS1 OS/VS2 R1.7 (SVS) TOS/360 TSS/360 VMF/370 this recent post references http://www.garlic.com/~lynn/2007u.html#18 Folklore references to CP67 at Lincoln Labs some (virtual machine) cp67 historical references from Melinda's VM paper at http://www.princeton.edu/~melinda/ mentioning that very early, two new commercial companies were formed to offer (virtual machine) cp67-based commercial timesharing services http://www.garlic.com/~lynn/subtopic.html#timeshare drawing people heavily from Science Center, http://www.garlic.com/~lynn/subtopic.html#545tech Lincoln Labs, and Union Carbide. It also makes references to MTS folklore having been initially built on top of Lincoln Labs LLMPS. There was an OS/360 operators console application called ONLINE/OS that provided CMS-like interactive functionality. It was most frequently used with PCP ... but could also be used on MFT and MVT. CP67 had a function that could save a virtual memory image of a running virtual machine. This was used with CMS to get rapid startup. However, a technique was developed that could also checkpoint a virtual memory image of OS/360 ... at point when I/O had been quiesed ... allowing OS/360 quick start in a virtual machine (just restore the saved virtual memory image). This could be used in conjunction with restoring a saved image of OS/360 where ONLINE/OS had already been up and running. old posts mentioning online/os http://www.garlic.com/~lynn/2001i.html#34 IBM OS Timeline? http://www.garlic.com/~lynn/2001n.html#45 Valid reference on lunar mission data being unreadable? http://www.garlic.com/~lynn/2004.html#48 AMD/Linux vs Intel/Microsoft http://www.garlic.com/~lynn/2004d.html#33 someone looking to donate IBM magazines and stuff http://www.garlic.com/~lynn/2007b.html#50 Is anyone still running part of Melinda's paper has appendix mentioning ONLINE/OS was never released outside the company (although I had a copy of it at the university in the 60s, also much of the original work had been done by a person on assignment from Union Carbide) ref: E.C. Hendricks, C.I. Johnson, R.D. Seawright, and D.B. Tuttle, Introduction to ONLINE/OS and ONLINE/OS User’s Guide, IBM Cambridge Scientific Center Reports 320-2036, 320-2037, March, 1969 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Open z architecture and Linux questions
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Wayne Driscoll) writes: Not really answering either question, but on the topic of Q2, the recent port of Open Solaris to System z was done only under z/VM, with no attempt to get it to run under LPAR mode because of the increased amount of work LPAR mode would have added (paraphrased from the company that did the porting work). Now that the hard part of getting Linux to run in an LPAR has been done, I don't see the need to eliminate it, but it would be interesting to see the percentage of Linux on z usage in LPAR vs z/VM. long ago and far away, similar arguments were made for both gold/au and aix/370. issue was that field engineering had lots of diagnostic, recording, and recovery requirements for servicing customer machines (EREP, RAS, etc). the effort to add mainframe EREP/RAS functionality to any of these ports was several times larger than just doing the straight forward port (while vm was able to satisfy the requirement, including for any of its guest operating systems). however, over the yrs, there has been more and more of virtual machine support functionality being moved into LPAR and service processor operation. slightly related recent post http://www.garlic.com/~lynn/2007t.html#77 T3 Sues IBM To Break its Mainframe Monopoly also in this post http://www.garlic.com/~lynn/2007u.html#8 Open z/Architecture or Not the reference to various OCO related material from vmshare archives, the reference to TUCC's MVS/370 to MVS/XA conversion experiences describes part of the success was having access to SIE and VM/SF information http://vm.marist.edu/~vmshare/browse?fn=OCOCMEft=NOTE ... part of difficulty discussion from above ... The key to gaining performance from the primary guest operating system is the I/O Passthru feature of SIE. This allows the guest system to initiate I/O directly to the I/O subsystem without intervention from VM/SF. The SIE microcode assist is a documented feature, however the portion that supports I/O Passthru is not documented. As a result it took us two months to correct this problem. The problem was extraordinarly difficult to analyze, because the symptoms were noticeable only after the problem occured. We had all of MVS/370's I/O devices in I/O Passthru, including the Memorex 1270 devices. In certain circumstances, such as MVS disabling for 09x wait, VM/SF decided to remove all of the I/O from I/O Passthru. After taking all devices out of I/O passthru, VM/SF will then put them all back in. Performing this function requires that VM/SF perform a Modify Subchannel to each device to accomplish this. ... snip .. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Crypto Related Posts
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: the x9.59 financial standard approach was then to fix the underlying weakness, lack of strong authentication ... which also then eliminated needing to hide the transaction information from crooks (since the information was useless w/o the proper authentication). some of this is discussed in the posts concerning the naked transaction metaphor http://www.garlic.com/~lynn/subintegrity.html#payments re: http://www.garlic.com/~lynn/2007t.html#61 Re: Crypto Related Posts some recent related: Why should merchants keep credit card data? http://www.networkworld.com/news/2007/120607-why-should-merchants-keep-credit.html the proposed approach was raised at least a decade ago ... it addresses harvesting data-at-rest in repostories ... but doesn't address the evesdropping and skimming attacks. http://www.garlic.com/~lynn/subtopic.html#harvest previous business process difficulties (with the suggested approach) was availability of online connectivity (giving merchants access to the necessary data for required/mandated business operations). the pervasive growth of internet connectivity has somewhat mitigated those issues. Can mid-market merchants comply with PCI standards? http://www.networkworld.com/news/2007/120607-can-mid-market-merchants-comply-with.html another approach that has been tried is the one time account numbers (as an approach to eliminating replay attacks ... aka eliminating being able to use information from previous transactions for fraudulent activity). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Jim Mulder) writes: But actually it did not take decades, as the original release of MVS/XA in 1982 functionally supported 16-way SMP. Of course there were no such processors at the time (nothing greater than 2-way until the 4-way 3084), but it did run for testing purposes using 16 virtual CPUs on a modified version of VM. Of course, as larger processors were actually built, additional was done (and continues to be done) to address performance/scaling issues. re: http://www.garlic.com/~lynn/2007t.html#76 T3 Sues IBM To Break its Mainframe Monopoly well, sort of. one of the things to get rapidly to 16-way smp implementation, as well as addressing performance/scaling issues, was to relax standard 370 cache consistency rules (and, in fact, most SMP vendors going to larger numbers of processors have almost always involved how to deal with cache consistency issues). remember that compareswap ... misc. posts about smp and/or compareswap http://www.garlic.com/~lynn/subtopic.html#smp was invented by charlie (compare-and-swap was chosen because CAS are charlie's initials) at the science center http://www.garlic.com/~lynn/subtopic.html#545tech and i've mentioned before the original difficulty of getting compare-and-swap into 370 architecture. Some of the difficulties is why the example of program failure still appears in the compare-and-swap writeup http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/DZ9ZR003/A.6.1?SHELF=DZ9ZBK03DT=20040504121320 i've frequently claimed that the 801 risc effort http://www.garlic.com/~lynn/subtopic.html#801 was attempt to go to the opposite extreme from what went on in FS http://www.garlic.com/~lynn/subtopic.html#futuresys and also claimed the lack of cache consistency in 801 risc was adverse reaction to the heavy performance penalty paid in 370 by its strong cache consistency requirement. in fact, it wasn't until somerset (joint ibm, motorola, apple, et all) for power/pc that there was (risc) work on smp and addressing cache consistency. in any case, part of doing 16-way smp (and relaxing 370 cache consistency rules) was much more detailed attention paid to every piece of code (because of the associated hardware changes for relaxed cache consistency). for some more topic drift, in just the 3084 time-frame, both mvs and (standard) vm had effort to go thru all kernel data storage management and make sure things were cache-line sensitised. the issue was the increased probability that more than one cache might be accessing different data items which happened to overlap in the same cache line (resulting in significant cache line thrashing). The claim at the time was that this effort resulted in 5-10 percent increased system thruput (for 4-way). As the number of independent caches that had to be coordinated ... the probability increases that there is going to be some kind of cache interference. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. re: http://www.garlic.com/~lynn/2007t.html#76 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007t.html#76 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007u.html#1 T3 Sues IBM To Break its Mainframe Monopoly for slightly more light-hearted, seasonal reference, old email with mvs/xa tso reference from long ago and far away: Date: 08/26/82 15:24:21 re: mvs/xa; i've seen it for myself, a 3081 system completely idle except for one MVS/XA tso user. Response time is longer for that single TSO user on the 3081 than for CMS doing same type of stuff on a loaded 3033. MVS/XA is copy of the one that large internal datacenter is using for their development work. the large internal datacenter has gen'ed the TSO logo screen (in big block letters) BAH HUMBUG The only thing slower than the 3081 service processer (5+ seconds to single step one instruction) on the 3081 is possibly MVS/XA TSO. The observation is that TSO is so slow, that you have lots of time to syntax your next input make sure that there are no mistakes (because if there are ... then things will really be slow). ... snip ... somewhat related to post in this thread http://www.garlic.com/~lynn/2007t.html#40 Why isn't OMVS command integrated with ISPF? -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Open z/Architecture or Not
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Bob Shannon) writes: Sure. The thousands of in-stream usermods that were written prior to XA, and which greatly inhibited subsequent upgrades. I certainly agree that in the early days usermods were written to overcome functional deficiencies in MVS. Some, such as logical swap, were incorporated into MVS. Others, such as the dual master catalog mod at a large US insurance company, proved to be a nightmare to maintain and an even worse nightmare to remove. cp67 and vm370 were notorious for user modifications ... in part because it shipped not only with full source ... but its whole customer maintenance infrastructure was source based (i.e. each fix shipped as incremental source update file). in the early 80s there was a study of local vm370 system modifications. internal corporate local modifications were as large as the base system ... and the share library source changes were approximately equivalent to the internal corporate local modifications (in size and function). part of all this started with unbundling announcement 23jun69 http://www.garlic.com/~lynn/subtopic.html#unbundle starting to charge for application software. however, the case was made that kernel code could still be free (bundled). A lot of the structural and functional enhancements that I had done to cp67 as an undergraduate (and was picked up and shipped in the product) was dropped in the morph from cp67 to vm370. However, I had done the port myself ... referenced in this prior post http://www.garlic.com/~lynn/2007t.html#69 T3 Sues IBM TO Break its Mainframe Monopoly and this old email http://www.garlic.com/~lynn/2006v.html#email731212 http://www.garlic.com/~lynn/2006w.html#email750102 I distributed and supported the CSC/VM system (mentioned in the above) for a large number of internal datacenters. The product picked up some small pieces of the above as part of VM370 rel3. However, other pieces were selected to be released as separate resource manager product ... and also got chosen to be guinea pig for unbundling/charging for kernel software (which met that i had to spend a lot of time with business people ironing about the policies for kernel software charging). http://www.garlic.com/~lynn/subtopic.html#fairwhare http://www.garlic.com/~lynn/subtopic.html#wsclock because of the extensive source oriented culture ... most customers managed to regularly track local source code changes as new releases came out. However, I know of (at least) one notable exception. Somehow or another, a very early CSC/VM system was leaked to ATT longlines. Over a period of years, they developed a large body of their own source changes ... never bothered to track releases, and migrated it to a number of their own machines. Nearly a decade later, I was tracked down by the ATT national marketing rep about trying to help get ATT longlines off this ancient CSC/VM system. The OCO-wars (object code only) in the early 80s were somewhat turbulent. There had been some number of commercial online timesharing services formed from cp67 and vm370. http://www.garlic.com/~lynn/subtopic.html#timeshare these were somewhat similar to the internal HONE systems that worldwide sales and marketing used http://www.garlic.com/~lynn/subtopic.html#hone One of these was Tymshare which in the mid-70s started providing the vmshare online discussion forum to share members. That vmshare forum has now been archived here http://vm.marist.edu/~vmshare/ included in the forum archives are the OCO-war discussions from the early 80s. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Open z/Architecture or Not
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: The OCO-wars (object code only) in the early 80s were somewhat turbulent. re: http://www.garlic.com/~lynn/2007u.html#6 Open z/Architecture or Not as before the vmshare archives are at http://vm.marist.edu/~vmshare/ old vmshare post about the original source maint infrastructure, originally developed on cp67 http://vm.marist.edu/~vmshare/read?fn=HISTORYft=MEMOline=49 a quicky search for some OCO related posts from archive ... this is discussion from 93 regarding the OCO's 10th b'day: http://vm.marist.edu/~vmshare/browse?fn=OCO:BDAYft=MEMO OCO Study Handouts from SHARE 72 (Feb89) http://vm.marist.edu/~vmshare/browse?fn=OCOSTUDYft=NOTE TUCC's MVS/370 to MVS/XA conversion experiences (Jun88) http://vm.marist.edu/~vmshare/browse?fn=OCOCMEft=NOTE VM Program Products which should be distributed with Source Code. (started May80) http://vm.marist.edu/~vmshare/browse?fn=VMSOURCEft=MEMO old email mentioning vmshare ... including discussing obtaining monthly copies of all vmshare files for putting up on the HONE system for worldwide sales and marketing http://www.garlic.com/~lynn/subtopic.html#hone and other internal systems. for other drift, one of the things i did during this period was do a rex(x)-implementation replacement for ipcs debugging tool. http://www.garlic.com/~lynn/subtopic.html#dumprx part of the issue was to demonstrate that rex(s) wasn't just another pretty exec language. the objective was to be able to replace the existing ipcs (which was a large body of assembler implemented code) with a 1) rex(x) implementation, 2) that took less than half-time over 3months to implement, 3) had ten times the function and 4) ten times the performance (took some slight of hand) a side-effect was that if it was decided to replace the existing implementation ... then source would have to be shipped for the new ipcs ... regardless of any OCO-policy. It was never decided to ship the implementation as replacement IPCS ... but it eventually came to be used at effectively all internal datacenters and the majority of PSRs processing customer reported problems. However, i was approved to give a share presentation on the implementation ... and within a couple months after the presentation, there were a number of similar implementations by various organizations. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Van Dalsen, Herbie) writes: And who came up with XA I/O? Amdahl, in order to do MDF and share channels had to do floating I/O interrupts, and related control block structures in HSA (a la XA) to get this to work. try 360/67 smp channel director for sharing channels and floating i/o interrupts ... 360/67 functional characteristics can be found here http://www.bitsavers.org/pdf/ibm/360/funcChar/ ... 360/67 had 24bit 32bit addressing modes, also referenced in this post http://www.garlic.com/~lynn/2007t.html#75 T3 Sues IBM To Break its Mainframe Monopoly after future system was killed http://www.garlic.com/~lynn/subtopic.html#futuresys there was mad rush to get out 303x in parallel with starting on xa. the architecture documents for xa, subchannel infrastructure, access registers, et all were referred to as 811 ... from their nov78 publication date (aka 29yrs ago). I had fairly complete copy ... they were individually numbered copies, classified at the highest level ... requiring special double-lock security filecabinet and periodic auditing. apparently information about people with copies leaked out and several people were approached ... aka industrial espionage ... and the feds eventually were involved. part of it involved the extrodinary lead time to move mvs to anything ... reference to killing vm370 product because they needed all the developers moved to pok to help meet mvs/xa delivery schedule http://www.garlic.com/~lynn/2007t.html#68 T3 Sues IBM To Break its Mainframe Monopoly even before 811 documents were published we had put together a project to turn out a 16-way smp processor on a very aggresive delivery schedule. it was going great guns until it came to the attention of the head of pok that it would possibly be decades before mvs ever had 16-way smp support (some people were then invited to never show up at the pok site again). misc. past posts mentioning smp support (and/or compare-and-swap instruction) http://www.garlic.com/~lynn/subtopic.html#smp there was small advanced technology conference in pok spring of 77 (a little over 30yrs ago) with presentations on both 16-way smp and 801 risc ... for lots of topic drift, misc. 801 risc posts http://www.garlic.com/~lynn/subtopic.html#801 misc. post posts mentioning 811 http://www.garlic.com/~lynn/2000d.html#21 S/360 development burnout? http://www.garlic.com/~lynn/2002d.html#8 Security Proportional to Risk (was: IBM Mainframe at home) http://www.garlic.com/~lynn/2002d.html#9 Security Proportional to Risk (was: IBM Mainframe at home) http://www.garlic.com/~lynn/2002d.html#49 Hardest Mistake in Comp Arch to Fix http://www.garlic.com/~lynn/2002d.html#51 Hardest Mistake in Comp Arch to Fix http://www.garlic.com/~lynn/2002j.html#28 ibm history note from vmshare http://www.garlic.com/~lynn/2002k.html#34 30th b'day original vm/370 announcement letter (by popular demand) http://www.garlic.com/~lynn/2002m.html#28 simple architecture machine instruction set http://www.garlic.com/~lynn/2002n.html#58 IBM S/370-168, 195, and 3033 http://www.garlic.com/~lynn/2003c.html#1 Wanted: Weird Programming Language http://www.garlic.com/~lynn/2004c.html#6 If the x86 ISA could be redone http://www.garlic.com/~lynn/2004g.html#24 |d|i|g|i|t|a|l| questions http://www.garlic.com/~lynn/2005j.html#34 IBM Plugs Big Iron to the College Crowd http://www.garlic.com/~lynn/2005j.html#35 IBM Plugs Big Iron to the College Crowd http://www.garlic.com/~lynn/2005p.html#18 address space http://www.garlic.com/~lynn/2005s.html#26 IEH/IEB/... names? http://www.garlic.com/~lynn/2006b.html#28 Multiple address spaces http://www.garlic.com/~lynn/2006f.html#20 Old PCs--environmental hazard http://www.garlic.com/~lynn/2006j.html#27 virtual memory http://www.garlic.com/~lynn/2006j.html#31 virtual memory http://www.garlic.com/~lynn/2006n.html#16 On the 370/165 and the 360/85 http://www.garlic.com/~lynn/2006n.html#27 sorting was: The System/360 Model 20 Wasn't As Bad As All That http://www.garlic.com/~lynn/2006t.html#16 Is the teaching of non-reentrant HLASM coding practices ever defensible? http://www.garlic.com/~lynn/2007g.html#57 IBM to the PCM market(the sky is falling!!!the sky is falling!!) http://www.garlic.com/~lynn/2007k.html#28 IBM 360 Model 20 Questions http://www.garlic.com/~lynn/2007l.html#71 IBM 360 Model 20 Questions misc. posts mentioning 16-way smp support http://www.garlic.com/~lynn/95.html#5 Who started RISC? (was: 64 bit Linux?) http://www.garlic.com/~lynn/95.html#6 801 http://www.garlic.com/~lynn/95.html#11 801 power/pc http://www.garlic.com/~lynn/97.html#5 360/44 (was Re: IBM 1130 (was Re: IBM 7090--used for business or http://www.garlic.com/~lynn/98.html#40 Comparison Cluster vs SMP? http://www.garlic.com/~lynn/2002i.html#82 HONE http://www.garlic.com/~lynn/2002p.html#58 AMP vs SMP
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Van Dalsen, Herbie) writes: And who came up with XA I/O? Amdahl, in order to do MDF and share channels had to do floating I/O interrupts, and related control block structures in HSA (a la XA) to get this to work. re: http://www.garlic.com/~lynn/2007t.html#75 T3 T3 Sues IBM To Break its Mainframe Monopoly http://www.garlic.com/~lynn/2007t.html#76 T3 T3 Sues IBM To Break its Mainframe Monopoly for other topic drift, a big part of the queued subchannel i/o interface was to compensate for the enormous mvs pathlength to (re)drive i/o ... lots of i/o idle between the end of the previous operation and initiating the next operation. part of this was also predicated that during the 70s, systems started to shift from being significantly processor constrained/bottlenecked to more and more being i/o bottlenecked. i had started pointing this out early ... and at one point some disk division executive assinged their performance group to refute the characterizations (i.e. over more than a decade, the relative disk system thruput had declined by an order of magnitude; aka disks got faster ... but other parts of systems had gotten an order of magnitude faster still). after some period they came back and pointed out that I had slightly understated the problem. this eventually turned into share presentation on how to optimize systems for disk thruput. the initial justification was that the queued interface allowd just moving the redrive operation from mvs kernel into the microcode of the same processor (not even offloaded to different processor), that the microcode engineers could do a significantly better redrive implementation that the mvs software developers. i had worked on a 5-way smp project in 75 where the processor complex had significant microcode capability ... some past posts http://www.garlic.com/~lynn/subtopic.html#bounce and i had defined a queued i/o interface ... but it included being able to offload much of it to a separate/decidated processor. i had also defined a queued microcode interface for dispatching ... allowing processors to pick off work w/o having to go thru the kernel function. this was canceled w/o shipping ... and some of the same people then reconstituted to work on 16-way smp effort mentioned in previous post. i was allowed to play disk engineer in bldgs. 1415 ... misc. posts http://www.garlic.com/~lynn/subtopic.html#disk and one of the things i worked on was the whole testcell testing infrastructure that was being done on stand-alone dedicated machines. They had tried MVS at one point with a single testcell but experienced 15mins MTBF (hangs, crashes, etc, requiring manual intervention and MVS reboot). I undertook to rewrite the i/o supervisor so that multiple testcells could be tested concurrently an the same machine in an operating system environment. This turned out to have very low processor utilization and so the engineers started also using the test machines for other purposes. bldg 15 got one of the first 3033 engineering machines (outside of POK) for disk testing. partly because things were going very well ... they also managed to put together 16 3330 disk drives and 3830 controllers where the machine could be concurrently used for other purposes. this was during a period when there was heavy 3880 controller development and testing going on. at one pointer there was a formal product performance acceptance test for the 3880 done in STL using standard operating system testing. then bright and early one monday i got a call from the engineers in bldg 15 asking what i had done over the weekend to totally destroy there system thruput. I said I hadn't done anything ... and they claimed they hadn't done anything. So i had to start diagnosing what went one. It turns out that over the weekend, they had replaced the 3830 (for the string of 16 3330 drives) with a 3880 controller. The problem was that in the move from 3830 to 3880 they went from a (fast) horizontal microcoded processor to a much slower vertical microcoded processor (with a separate data path). As a result, the 3880 had much slower command and funtion processing ... and initially failed the formal product performance acceptance test. The 3880 was then tweaked to present early interrupt to the channel (indicating operation complete) before the 3880 had finished all its operation. Then the 3880 could complete its operation in parallel with the operating system processing the interrupt and getting around to redriving i/o. This didn't bother the standard operating system formal performance acceptance tests. The problem was that I had significantly redone the I/O subsystem, not only to make it much more reliable and available than standard MVS ... but interrupt processing was dramatically faster than standard MVS ... and would get around to redriving i/o
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Roger Bowler) writes: IBM worked long and hard over many years to successfully establish S/360 and its successors as *the* standard computer architecture. Indeed for a 20 year period between about 1970 to 1990 S/360/370/390 was almost the only architecture which would reasonably be considered for most business systems large or small. With the result that applications tied to MVS and VSE are now firmly embedded into the infrastructure of the various information systems (banks, utilities, government, airlines) that allow our society to function the way it does. The figure of $1 trillion invested in software compatible with IBM mainframes has been widely quoted. some recent topic drift in thread that wandered into run-up/justification for 360 http://www.garlic.com/~lynn/2007t.html#63 Remembering the CDC 6600 http://www.garlic.com/~lynn/2007t.html#65 Remembering the CDC 6600 also referenced in the above, there was an failed/aborted attempt to take a large detour in the early 70s with the future system effort http://www.garlic.com/~lynn/subtopic.html#futuresys motivated by the growth in the plug-compatible controller business ... discussed in more detail in this recent post http://www.garlic.com/~lynn/2007r.html#74 System 360 EBCDIC vs. ASCII it was in the FS period that Amdahl launched his plug-compatible processor business. In the early 70s, Amdahl gave a talk at MIT where he was quized about it. One of the questions was what justification did he use to raise funding for the company. The response was something about customers had already spent $200b in 360-based application software, and even if IBM were to totally walk away from 360 (could possibly be considered a veiled reference to the future system project), that software base would be sufficient to keep him in business through the end of the century (i.e. the $200b number was less than a decade after 360 had been announced). The future system distraction drew a lot of resources away from 370 activities. When future system was finally killed, there was mad scramble to get software and hardware products back into the 370 product pipeline. The lack of products in the 370 product pipeline possibly contributed to market opportunities for clone processor vendors. The 303x were part of that mad scramble ... which was effectively started in parallel with what was to become 3081 and 370-xa. recent post going into details of 303x effort http://www.garlic.com/~lynn/2007p.html#1 what does xp do when system is copying however, it was the long lead-time to do mvs/xa and the associated mad scramble that led to justification to kill vm370 and transfer everybody from the burlington mall vm370 group to pok ... supposedly as necessary in order to meet the mvs/xa schedule. endicott eventually did manage to acquire the vm370 product mission and keep it alive ... but effectively had to reconstitute the group from scratch. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: some recent topic drift in thread that wandered into run-up/justification for 360 http://www.garlic.com/~lynn/2007t.html#63 Remembering the CDC 6600 http://www.garlic.com/~lynn/2007t.html#65 Remembering the CDC 6600 also referenced in the above, there was an failed/aborted attempt to take a large detour in the early 70s with the future system effort http://www.garlic.com/~lynn/subtopic.html#futuresys motivated by the growth in the plug-compatible controller business ... discussed in more detail in this recent post http://www.garlic.com/~lynn/2007r.html#74 System 360 EBCDIC vs. ASCII re: http://www.garlic.com/~lynn/2007t.html#68 T3 Sues IBM To Break its Mainframe Monopoly i wasn't exactly unbiased ... i had somewhat ridiculed the future system effort during the period (drawing comparison with a cult movie that had been playing down in central sq) and continued to work on 370 stuff (including making statements about the resource manager, that i already had running, was better than the theoritical pipe dreams being specified in future system architecture documents). slightly related old email from the period http://www.garlic.com/~lynn/2006v.html#email731212 http://www.garlic.com/~lynn/2006w.html#email750102 http://www.garlic.com/~lynn/2006w.html#email750430 http://www.garlic.com/~lynn/2006w.html#email750827 in the mad rush, after FS was killed, contributed to decisions to pickup some of the work (that i had continued to do) and ship it in products. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: T3 Sues IBM To Break its Mainframe Monopoly
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Lindy Mayfield) writes: I was thinking (dreaming) today about what if when I giving training for MVS stuff and each student had their own mainframe instead of connecting to a central one. We could do so much more. This was somewhat the original idea behind HONE (hands-on network environment) http://www.garlic.com/~lynn/subtopic.html#hone after the 23jun69 unbundling announcement http://www.garlic.com/~lynn/subtopic.html#unbundle Prior to unbundling, new/young system engineers acquired quite a bit of their knowledge, working as part of team at customer installations. the unbundling announcement pretty much put an end to this apprentice-like activity. In the 60s, while an undergraduate at the univ, i was doing a large number of os/360 enhancements ... and prior to unbundling, they would cycle new SEs thru the univ. every six months (that I would get to train). HONE started out creating a number of cp67 virtual machine datacenters that would support branch office system engineers running (guest) operating systems. The cp67 virtual machine systems (running on real 360/67s) included enhancements supporting simulation of the newly announced (pre-virtual memory) 370 instructions (allowing newer guest operating system versions to be run). The science center http://www.garlic.com/~lynn/subtopic.html#545tech had also ported apl\360 to cms\apl and reworked it for operation in a virtual memory environment (including arbitrarily large workspaces, up to 16mbytes, rather than the somewhat toy apl\360 workspaces that were typically 16kbytes to 32kbytes). CMS\APL on HONE was leveraged to also deploy a large number of sales and marketing support applications. These applications soon dominated HONE utilization and the original HONE purpose somewhat withered away. For example, by the mid-70s, it was no longer possible to submit a customer order that hadn't first been processed by HONE configurators and/or other applications (although by this time, HONE had migrated to VM370 and APL\CMS). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Crypto Related Posts
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Patrick O'Keefe) writes: We've been moving towards secure data transmissions with business partners (with testing starting in about a week) but nobody had checked what we could do to lessen the effect encryption would have on our processors. the internal network http://www.garlic.com/~lynn/subnetwork.html#internalnet required encryption on all traffic that left corporate facility (like inter-site traffic). when most places were doing 9.6kbit, we were working on full-duplex T1 ... and looking at associated problems in HSDT project http://www.garlic.com/~lynn/subnetwork.html#hsdt old email reference looking at problem supporting full duplex T1 on 3081 http://www.garlic.com/~lynn/2006n.html#email841115 in this post discussing encryption technology http://www.garlic.com/~lynn/2006n.html#36 The very first text editor part of the effort was looking at being able to move past hardware link encryptors as solution to the opportunities (which was the default state-of-the-art at the time). other old crypto related email http://www.garlic.com/~lynn/lhwemail.html#crypto included old email reference to a non-certificate-based public key infrastructure for more secure communication over the network http://www.garlic.com/~lynn/2006w.html#email810515 in this post: http://www.garlic.com/~lynn/2006w.html#12 more secure communication over the network other recent posts to discontinuity between the hsdt project (doing T1 and faster interconnects) while other operations were still looking at 9.6kbit with a little more to (high-speed) 56kbit http://www.garlic.com/~lynn/2007q.html#45 Are there tasks that don't play by WLM's rules for other old topic drift ... we had been called to come in and consult with this small client/server startup that wanted to do payments on their server http://www.garlic.com/~lynn/subnetwork.html#gateway and they had this technology they invented called SSL that they wanted to use as part of the implementation. we had to do some end-to-end studies looking at how to apply the technology to the business processes some related posts http://www.garlic.com/~lynn/subpubkey.html#sslcerts afterwards we participated in the x9a10 financial standard working group that in the mid-90s had been given the requirement to preserve the integrity of the financial infrastructure for all retail payments. the result was the x9.59 financial standard http://www.garlic.com/~lynn/x959.html#x959 x9.59 took a slightly different approach, in part because of detailed end-to-end threat and vulnerability analysis involving retail payments, basically identifying significant authentication vulnerability which then led to requirements for hiding transaction information ... as countermeasure to crooks obtaining the information and being able to perform fraudulent transactions. the observation was that the transaction information was needed at a large number of different processes, potentially occuring over extended period of time ... and as a result, even if the planet was buried under miles of crypto ... it still wouldn't prevent information leakage. the x9.59 financial standard approach was then to fix the underlying weakness, lack of strong authentication ... which also then eliminated needing to hide the transaction information from crooks (since the information was useless w/o the proper authentication). some of this is discussed in the posts concerning the naked transaction metaphor http://www.garlic.com/~lynn/subintegrity.html#payments -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Data Center Theft
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Timothy Sipples) writes: In fairness, the DS6000 is physically relatively small, although I wouldn't want to carry one by myself on my bicycle. The spindles (individual drives) are even smaller, but you'd need a number of them to have a RAID set and the complete data. Tough but not impossible. I think the IT marketplace is in for a shock when people figure out that losing the keys means losing the data. It isn't like a bank vault where you can hire a locksmith to drill some holes over several days. It's so critical to store and manage the encryption keys in a safe, secure, recoverable repository. can you say key escrow? ... this was one of the themes from the key escrow meetings from the mid-90s. however, there was lot of confusion about what key escrow met, i.e. 1) gov. held all keys? 2) institutions holding keys for their own data encryption (as an availability, business continuity and no-single-point-of-failure)? 3) all kinds of keys?, authentication as well as encryption 1 got lots of bad press including all the swirl around clipper chip and things like LEAF 3 authentication keys aren't really an availability issue ... and could violate some basic security principles regarding being able to associate all activities uniquely with individuals. with all the bad press ... various key escrow activities sort of just evaporated wiki reference: http://en.wikipedia.org/wiki/Key_escrow nist references http://csrc.nist.gov/keyrecovery/ misc. past posts mentioning key escrow http://www.garlic.com/~lynn/aadsm9.htm#pkcs12 A PKI Question: PKCS11- PKCS12 http://www.garlic.com/~lynn/aadsm16.htm#11 Difference between TCPA-Hardware and a smart card (was: example: secure computing kernel needed) http://www.garlic.com/~lynn/aadsm18.htm#12 dual-use digital signature vulnerability http://www.garlic.com/~lynn/aadsm23.htm#6 PGP master keys http://www.garlic.com/~lynn/2001c.html#65 Key Recovery System/Product http://www.garlic.com/~lynn/2001h.html#7 PKI/Digital signature doesn't work http://www.garlic.com/~lynn/2001i.html#36 Net banking, is it safe??? http://www.garlic.com/~lynn/2001j.html#52 Are client certificates really secure? http://www.garlic.com/~lynn/2002d.html#39 PKI Implementation http://www.garlic.com/~lynn/2003j.html#53 public key confusion http://www.garlic.com/~lynn/2004i.html#12 New Method for Authenticated Public Key Exchange without Digital Certificates http://www.garlic.com/~lynn/2006d.html#39 transputers again was Re: The demise of Commodore http://www.garlic.com/~lynn/2006d.html#40 transputers again was Re: The demise of Commodore http://www.garlic.com/~lynn/2007c.html#1 Decoding the encryption puzzle -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Why isn't OMVS command integrated with ISPF?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Paul Gilmartin) writes: Anyone who believe that's a fundamental limitation of 3270 hardware that can't be worked around: o Has never used VM/CMS o Has been brainwashed by TSO ... probably both. On CMS, I can type input to my program, while it runs, in anticipation of a VM READ. I can type immediate commands to my Rexx EXEC to turn on tracing with no ATTN nor need to wait for in input prompt. we actually did some hardware mods to 3277 to eliminate race condition if you happened to type at the instant the system wrote to the terminal (which would lock the keyboard) ... aka 327x being half-duplex infrastructure. we complained about the change-over to 3274 controller with 3278 terminal (i.e. effectively terminal manufacturing cost reduction moving a lot of components back into shared controller). having shared electronics back in 3274 controller made 3278 terminal operations (including response) a lot slower. complaining about it basically got a response was that the significant hardware slowdown effectively wasn't noticeable since mvs (tso) was so slow anyway that it wasn't noticeable. post with old 3272/3274 comparisons http://www.garlic.com/~lynn/2001m.html#19 3270 protocol difference als shows up later with terminal emulation and the difference between file download with ANR (i.e. 3272/3277) and DCA (i.e. 3274/3278) protocols (anr three times dca thruput) lots of past posts mentioning terminal emulation http://www.garlic.com/~lynn/subnetwork.html#emulation for some total topic drift ... old email mentioning tso product manager asking me if i would consider doing version of my resource manager for mvs/tso operation (this was after marketing division decided to start marketing CMS as the corporations strategic interactive product) http://www.garlic.com/~lynn/2006b.html#email800310 http://www.garlic.com/~lynn/2006v.html#email800310b reference in these posts http://www.garlic.com/~lynn/2006b.html#39 another blast from the past http://www.garlic.com/~lynn/2006v.html#23 Ranking of non-IBM mainframe builders? in some sense, CMS provided interactive personal computing in 60s, 70s and some part of the 80s ... but then saw personal computing starting to shift to PCs. for other folklore topic drift, cern did a report at share circa '74 about tso/cms bakeoff. internally within the company, copies of the report were classified confidential - restricted (i.e. available on need-to-know only) ... aka while they couldn't restrict its availability to customers ... they could restrict its availability to people in marketing and product development. ... and courtesy of the science center http://www.garlic.com/~lynn/subtopic.html#545tech virtual machines, cp67, cms (originally stood for cambridge monitor system before renamed to conversational monitor system as part of vm370 morph), gml (invented in '69 at the science center) precursor to sgml, html, xml, etc http://www.garlic.com/~lynn/subtopic.html#sgml and internal network technology http://www.garlic.com/~lynn/subnetwork.html#internalnet also used in bitnet/earn http://www.garlic.com/~lynn/subnetwork.html#bitnet here is reference discussing transformation from sgml to html at cern http://infomesh.net/html/history/early and first webserver outside europe was on slac vm370 system: http://www.slac.stanford.edu/history/earlyweb/history.shtml -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: SMF Under VM
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Kopischke, David G.) writes: From what I understand, we just use canned processes to extract SMF, load databases and create reports. But since we don't use VM at home and have no experience with it, maybe we're just not understanding where this data is in that environment ??? Is there any documentation that specifies what SMF data is available under VM and what is not ??? With respect to %CPU BUSY, I understand it's virtual under VM, but there still has to be some method of gauging how much CPU a guest is using, isn't there ??? How do VM shops report this ??? for decades VM would account for processor useage (both virtual and total) which would turned out to correspond very closely with total/actual busy (which was also measured). other infrastructures have tended to have accounted for processor busy which has been less than total/actual cpu busy (measured by other methods). The difference (which has peridically been quite substantial) was frequently referred to as capture ratio ... aka the sometimes small percentage of cpu busy that was actually accounted for for some, the concept of capture ratio took quite a bit of time to sink thru ... since a system not accounting for all cpu useage was quite foreign concept. running under VM ... one possible way for handling the (captured ... at least by vm) non-virtual processing time might be handled along with all the other uncaptured processor time (from the standpoint of a guest operating system running in a virtual machine). some of this also has to be handled with LPARs w/o VM software ... since LPARs are essentially a stripped down VM subset moved into the microcode of the machine (and then you can have virtual guests running in a VM software virtual machine ... which, in turn might be running in a LPAR virtual machine ... which is finally running on the real hardware). a few results for quicky search engine use for term cature ratio http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/FLASH10526 http://publib.boulder.ibm.com/tividd/td/TDS390/SH19-6818-08/en_US/HTML/DRLM9mst48.htm http://www.ibm.com/developerworks/wikis/display/zosperfinstr/Controlling+SMF+Record+Production http://www.ibm.com/developerworks/websphere/library/techarticles/0407_garza/0407_garza.html http://www.cmg.org/measureit/issues/mit38/m_38_10.html the original cp67 system delivered to the univ. the last week of jan68 did have something slightly reminiscence of uncaptured ... which was actual captured (i.e. specifically measured processor time) that wasn't associated with any specific operation (called overhead). This would increase significantly as the number of concurrent processes increased (aka it scaled extremely poorly). I completely reworked that implementation to eliminate the non-scaling characteristic ... as well as being able to account for what was actually being done. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Running REXX program in a batch job
The following message is a courtesy copy of an article that has been posted to alt.folklore.computers as well. [EMAIL PROTECTED] (Paul Gilmartin) writes: I believe Cowlishaw's book reports that Rexx was developed in the VM and MVS environments concurrently. It flourished in the former and withered in the latter, less likely because CLIST fulfilled the need better than EXEC2 than because less enthusiasm for innovation exists in the MVS environment (case in point: TCP/IP). Rexx didn't resurface under MVS until TSO/E. well, not quite: posting of old reference/quote http://www.garlic.com/~lynn/2005j.html#41 TSO replacement from Melinda's vm history paper http://www.princeton.edu/~melinda/25paper.pdf one of the quotes: Mike Cowlishaw had made the decision to write a new CMS executor on March 20, 1979. two months later, he began circulating the first implementation of the new language, which was then called ``REX''. Once Mike made REX available over VNET, users spontaneously formed the REX Language Committee, which Mike consulted before making further enhancements to the language. He was deluged with feedback from REX users, to the extent of about 350 mail files a day. By consulting with the Committee to decide which of the suggestions should be implemented, he rather quickly created a monumentally successful piece of software. ... snip ... similar comments are mentioned in rexx wiki page http://en.wikipedia.org/wiki/REXX and another old quote from this post http://www.garlic.com/~lynn/2006p.html#31 25th Anniversary of the Personal Computer from one of the references in the above: By far the most important influence on the development of Rexx was the availability of the IBM electronic network, called VNET. In 1979, more than three hundred of IBM's mainframe computers, mostly running the Virtual Machine/370 (VM) operating system, were linked by VNET. This store-and-forward network allowed very rapid exchange of messages (chat) and e-mail, and reliable distribution of software. It made it possible to design, develop, and distribute Rexx and its first implementation from one country (the UK) even though most of its users were five to eight time zones distant, in the USA. ... snip ... and for other topic drift, this was part of the thread related to the internal network http://www.garlic.com/~lynn/subnetwork.html#internalnet having been larger than the arpanet/internet from just about the beginning until sometime mid-85. part of the internal network issues was that while there were mvs/jes2 nodes, nearly all the nodes were vm ... and the jes2 nodes had to be carefully regulated. there were several issues around why jes2 nodes had to be carefully regulated on the internal network (some independent of the fact that the number of vm systems were significantly larger than the number of mvs systems) 1) jes2 networking started out being some HASP mods from TUCC that defined network nodes using the HASP psuedo device table ... limited to 255 entries ... 60-80 entries nominally taken up by psuedo spool devices ... leaving possibly only 170 entries for network node definitions 2) jes2 implementation would discard traffic if it didn't have either the origin or destination node in local defintion. the internal network had more nodes than jes2 could define for the majority of its lifetime ... so jes2 needed to be restricted to boundary nodes (at least not discarding traffic just passing thru). 3) jes2 implementation had a number of other deficiencies, including having confused header information as to network specific and local process handling. different versions or releases with minor variation in headers would bring down whole mvs system. even restricted to purely boundary nodes, there is infamous story of jes2 upgrade in san jose resulted in mvs system crashes in hursley. as a consequence there was special vm drivers created for talking to mvs jes2 systems ... which would convert jes2 headers to compatible format for the specific system on the other end of the line. this was somewhat the side-effect of the vm implementation having separated networking control information from other types of information ... effectively providing a kind of gateway implementation ... something not possible in the JES2 networking infrastructure (including not having a way from protecting itself from other JES2 systems, requiring intermediary vm systems to keep the JES2 systems from crashing each other). misc. past posts mentioning hasp, jes2, and/or jes2 networking http://www.garlic.com/~lynn/subtopic.html#hasp at some point ... while VM could run native protocol drivers as well as (multiple different) JES2 drivers ... JES2 could only run a specific JES2 drivers ... it was decided to start shipping VM only with JES2 drivers (even tho the native VM protocol drivers were more efficient). this was seen in the bitnet deployment http://www.garlic.com/~lynn/subnetwork.html#bitnet the vm tcp/ip product was developed
Re: CSA 'above the bar'
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Shmuel Metz , Seymour J.) writes: PSA is real address 0; it's absolute address 0 for at most one of the processors in the complex. Neither real nor absolute addresses are virtual addresses, and the mapping of virtual 0 to real 0[1] is strictly a software convention. multiprocessor support required a unique PSA for every processor. in 360 multiprocessor, the prefix register (for every processor) contained the real address of the PSA for that processor; different processors chose different real addresses for their PSA ... so as to have a unique PSA for every processor in the complex. the real, real page zero was no longer addressable (assuming every processor chose some other real address than zero). this was modified for 370, the prefix register specified the real address of the PSA for that processor (as in 360) ... however, if the processor addressed the address in the prefix register, it would reverse translate to real page zero. as a result, the real, real page zero could be used as common communication area between all processors in the complex ... and was addressed by using the address in the processor's prefix register. http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/dz9zr003/3.7?DT=20040504121320 from above: 1. Bits 0-50 of the address, if all zeros, are replaced with bits 0-50 of the prefix. 2. Bits 0-50 of the address, if equal to bits 0-50 of the prefix, are replaced with zeros. 3. Bits 0-50 of the address, if not all zeros and not equal to bits 0-50 of the prefix, remain unchanged. ... snip ... #1 #3 was how things operated in 360 multiprocessor support; #2 was introduced with 370 multiprocessor support (modulo 360 370 were 4kbyte pages and above description is for 64bit Z and 8kbyte pages) past posts mentioning multiprocessor support and/or compareswap instruction http:/www.garlic.com/~lynn/subtopic.html#smp other recent posts in this thread: http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#62 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#64 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#65 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#67 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#69 CSA 'above the bar' -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: [ClassicMainframes] multics source is now open
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Farley, Peter x23353) writes: Thanks a lot for the info and the link. Most interesting. Another important piece of computer history available to the world at large. Bravo to Bull for releasing it. It would be an interesting project to write the emulator for that machine architecture. for even more topic drift ... recent RDBMS related post ... drifting into mentioning that Multics shipped the first RDBMS product, MRDS http://www.garlic.com/~lynn/2007s.html#20 Ellison Looks Back As Oracle Turns 30 and even further drift in followup http://www.garlic.com/~lynn/2007s.html#21 Ellison Looks Back As Oracle Turns 30 also mentions that Multics unbundled MRDS ... which created issues about not allowing base system (free) software to have dependencies on priced software. lots of past posts mentioning RDBMS http://www.garlic.com/~lynn/subtopic.html#systemr I ran into a similar problem when my resource manager was selected to be guinea pig for starting to charge for kernel software. I had a lot of kernel restructuring (in the resource manager) for multiprocessor operation. This created a problem for bundled/free multiprocessor operation needed lots of code from the resource manager. http://www.garlic.com/~lynn/subtopic.html#unbundle -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: ATMs
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (R.S.) writes: Yes, I can. AFAIK z/OS version is not popular one. I know *big* ATM installation which migrated from z/OS to NonStop. People from ACI claimed that most of their installtions are not on mainframe. Timothy: I like mainframes, I have personal interest in mainframe business growth (at least survive), but I see no reason to be unhonest. a reference from hp/nonstop ACI’s BASE24 on the NonStop server hits 40 billion transaction mark http://www.hp.com/products1/24x7/strategic/aci.html disclaimer ... we did some marketing against them when we were doing our ha/cmp product ... misc. past posts http://www.garlic.com/~lynn/subtopic.html#hacmp for even more drift, latest newsletter http://www.tandemworld.net/newsletter%20nov07.htm and in later life, even worked on some joint projects with ACI. for instance AADS work http://www.garlic.com/~lynn/x959.html#aads nacha AADS rfi (submitted on our behalf, since were weren't nacach members) http://www.garlic.com/~lynn/nacharfi.htm involved modifying pre-auth capability in the EFT (debit) network switch pilot results http://internetcouncil.nacha.org/docs/ISAP_Pilot/ISAPresultsDocument-Final-2.PDF -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: ATMs
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Timothy Sipples) writes: It's an interesting bit of history that the first Tandem machine wasn't available until 1976, well after the first electronic ATM (1967) and lots of other ATMs. From what I've read the first networked ATM appeared in 1968, and the first popular ATM (i.e. same model placed into service by more than one bank) was the IBM 2984 starting in 1973. The IBM 2984 offered variable cash withdrawals and instantly deducted from your account, so it was 100% on-line -- 34 years ago. (I remember my father using our local bank's first ATM, newly installed, when I was a young child. It seemed like magic.) Presumably most if not all of these ATMs connected to IBM System/360s and /370s. Tandem came along after almost a decade of ATMs. re: http://www.garlic.com/~lynn/2007s.html#6 ATMs early work was done at los gatos lab ... before i was spending any time there. however, i do remember people talking about having worked on the development. they had large supply of bills from numerous different countries ... which they kept in a locked vault in the basement (for testing with the machines during development). they also mentioned story about one of the early machines going in across the street from a fast food resturant and kids feeding condiment packets into the card slot (one of the early bug fixes was countermeasure for such an attack). old posts reference 2984 http://www.garlic.com/~lynn/2006q.html#5 Materiel and graft http://www.garlic.com/~lynn/2006u.html#40 New attacks on the financial PIN processing http://www.garlic.com/~lynn/2006x.html#9 Plurals and language confusion http://www.garlic.com/~lynn/2007l.html#47 My Dream PC -- Chip-Based -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Poster of computer hardware events?
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (, IBM Mainframe Discussion List) writes: I made a mistake. A track not in the cache would take on the order of 20 milliseconds, so that would equate to 20 days instead of one day. A track already cached would result in an access time of one millisecond. If the 4K block can be found in a buffer somewhere in virtual storage inside the processor, it might take from 100 to 1000 instructions to find and access that data, which would equate to 100 to 1000 seconds, or roughly one to 17 minutes. And that assumes that the page containing the 4K block of data can be accessed without a page fault resulting in a page-in operation (another I/O), in which case we are back to several days to do the I/O. By the way, it takes at least 5000 instructions in z/OS to start and finish one I/O operation, so you can add about two hours of overhead to perform the I/O that lasts for 20 days. You really want to avoid doing an I/O if at all possible. reply to comment about RPS-miss (in the vmesa-l flavor of this thread) http://www.garlic.com/~lynn/2007s.html#5 Poster of computer hardware events? i had been making comments over a period of yrs that disk relative system thruput had declined by an order of magnitude (i.e. disks were getting faster but processors were getting much faster, faster). this eventually led to somebody in the disk division (gpd) to assigning the gpd performance group to refute the statements. after several weeks they came back and effectively said that i had somewhat understated the disk relative system thruput degradation ... when RPS-miss was taken into account. they then put a somewhat more positive spin on it and turned it into share 63 presentation b874 ... some past references: http://www.garlic.com/~lynn/2002i.html#18 AS/400 and MVS - clarification please http://www.garlic.com/~lynn/2006f.html#3 using 3390 mod-9s http://www.garlic.com/~lynn/2006o.html#68 DASD Response Time (on antique 3390?) one of the issues is does the 5k instruction pathlength roundtrip from EXCP (including channel program translation overhead) or roundtrip just after it has been passed to i/o supervisor??? for comparison numbers ... i had gotten cp67 total roundtrip for page fault down to approx. 500 instructions ... this included page fault handling, page replacement algorithm, a prorated fraction of page i/o write pathlength (which includes everything to start/finish i/o), total page i/o read pathlength (including full i/o supervisor), and two task switches thru dispatcher (one to switch to somebody else, waiting on the page fault to finish and another to switch back after the page i/o read finishes). to get it to 500 instructions involved touch almost every piece of code involved in all of the operations. I believe the 5000 instruction number was one of the reasons that 3090 extended store was a syncronous instruction (since the asyncronous overhead and all related gorp in mvs was so large). earlier, there had been some number of electronic 2305 paging device deployed at internal datacenters ... referred to as 1655 model (from an outside vendor). these involved effectively low latency but limited to channel transfer and cost whatever the asyncronous processing overhead. the 3090 extended store was done because of physical packaging issues ... but later when physical packaging was no longer an issue ... there were periodic discussions about configuring portions of regular memory as simulated extended store ... to compensate for various shortcomings in page replacement algorithms. with regard to the cp67 500 instruction number vis-a-vis MVS ... i would periodically take some heat regarding MVS having much more robust error recovery as part of the 5000 number (even tho the 500 number was doing significantly more). so later when i was getting to play in bldgs 14 15 (dasd engineering lab and dasd product test lab), i had opportunity to rewrite vm370 i/o supervisor. the labs in bldg. 1415 were running processor stand-alone testing for the dasd/controller testcells (one at a time). They had tried doing this under MVS but had experienced 15min MTBF (system crashing and/or hanging with just a single testcell). I undertook to completely rewrite i/o supervisor to make it absolutely bullet proof, allow concurrent testcell operation in operating system environment. lots of past posts mentioning getting to play disk engineer http://www.garlic.com/~lynn/subtopic.html#disk some old postings about comparisons of degradation of disk relative system thruput. the claim was that doing similar type of cms workload ... in going from cp67 on 360/67 with 80 users to vm370 on 3081 ... it should have shown an increase to several thousand online uses ... instead of increase to 300 or so online users. The increase in online users is roughly the
Re: Real storage usage - a quick question
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Knutson, Sam) writes: You should have the PTFs for z/OS APAR OA17114 installed if you are using paged fixed buffers in DB2 V8. Not having it was one of the causes of a z/OS outage here when a DB2 DBA accidently overcommitted storage to DB2. aka application page fixed buffers ... allows applications to specify the real addresses in the channel program ... avoiding the dynamic channel program translation (creating a duplicate of the channel program passed by excp/svc0) and dynamic page fixing that otherwise has to occur on every i/o operations (however, it can eliminate pageable storage needed by the rest of system) recent post mentioning difference between EXCP and EXCPVR (vis-a-vis channel program translation) http://www.garlic.com/~lynn/2007q.html#8 GETMAIN/FREEMAIN and virtual storage backing up other recent posts discussing dynamic channel program translation (in the initial translation from MVT to OS/VS2 supporting virtual memory, there was extensive borrowing of technology from cp67 CCWTRANS, channel program translation) http://www.garlic.com/~lynn/2007e.html#19 Cycles per ASM instruction http://www.garlic.com/~lynn/2007e.html#27 IBM S/360 series operating systems history http://www.garlic.com/~lynn/2007e.html#46 FBA rant http://www.garlic.com/~lynn/2007f.html#0 FBA rant http://www.garlic.com/~lynn/2007f.html#6 IBM S/360 series operating systems history http://www.garlic.com/~lynn/2007f.html#33 Historical curiosity question http://www.garlic.com/~lynn/2007f.html#34 Historical curiosity question http://www.garlic.com/~lynn/2007k.html#26 user level TCP implementation http://www.garlic.com/~lynn/2007n.html#35 IBM obsoleting mainframe hardware http://www.garlic.com/~lynn/2007o.html#37 Each CPU usage http://www.garlic.com/~lynn/2007o.html#41 Virtual Storage implementation http://www.garlic.com/~lynn/2007p.html#69 GETMAIN/FREEMAIN and virtual storage backing up http://www.garlic.com/~lynn/2007p.html#70 GETMAIN/FREEMAIN and virtual storage backing up http://www.garlic.com/~lynn/2007p.html#72 A question for the Wheelers - Diagnose instruction http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Real storage usage - a quick question
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Veilleux, Jon L) writes: In z/OS 1.8 the memory management is much more conducive to large memory. They no longer use the least recently used algorithm and no longer check every page. This has made a big difference for us. Under 1.7 we had issues with large real memory sizes due to the constant checking by RSM. This is no longer the case and we have increased our memory dramatically with no performance hit. one of the things found in clock LRU-approximation that i had originally done as undergraduate in the 60s http://www.garlic.com/~lynn/subtopic.html#wsclock was that if the interval between page resets started to exceed some limit, then there was little differention benefit of the reset activity ... least recently used tends to have some implicit dependencies on amount of history ... if the duration is too long ... then it lost much of its correlation being able to differentate between pages as to future page reference pattern. however across a wide range of configurations and workloads in the 70s, clock LRU-approximation had the advantage of effectively being able to (usefully) dynamically adapt the interval. however with a lot of cp67 experimenting and also heavy use of storage reference traces and page replacement modeling ... it was possible to show that outside some useful operating range ... the use of LRU algorithms for differentiating/predicting future page reference behavior became less and less accurate. It was also possible to show that for very large memories ... that the overhead of repeatedly resetting page reference bits provided less benefit than any possible improvement in page replacement strategy. we did do some experimenting at the science center attempting to recognize the operating region/environment across where clock LRU-approximated was beneficial ... and attempt to take some secondary measures/strategies when it was outside that operating region/envrionment. one of the scenarios was that most LRU-approximation algorithms are measured against how well they performed vis-a-vis simulation that exactly implemented least-recently-used page ordering (measured in terms of total page faults for given workload and real storage size). Good approximations tended to come within 5-15 percent (total page faults) of real least-recently-used page ordering. We were able to find some page replacement variations that instead of being 5-15 percent worse/more (total page faults compared to simulated real least-recently-used page ordering), we were able to show 5-15 percent fewer total page faults. the scenario was that in some configuration/workload scenarios, LRU-approximate could effectively cycle thru every page in real storage w/o finding a candidate ... and then take the first page it started with. Besides having a lot of processing overhead, this characteristic effectively degraded to FIFO page replacement (there are operating regions for LRU where it can degenerate to FIFO page replacement at the same time taking an extrodinary amount of processor overhead). our variation tended to recognize when operating in this configuration/workload region and effectively switched to RANDOM page replacement at very low processor overhead (and modeling showed that when not able to make any other differentiation between pages to be replaced ... RANDOM replacement makes better choice than FIFO, independent of the overhead issue). In fact, the original cp67 delivered at the univ. last week jan68, ... also referenced here http://www.garlic.com/~lynn/2007r.html#74 System 360 EBCDIC vs. ASCII ... effectively implemented something that tended to operate as FIFO replacement with purely software and didn't make use of the hardware reference bits. As undergraduate, I did the kernel algorithm and software changes to implement clock LRU-approximation page replacement ... taking advantage of page replacement bits. In this scenario ... with only on the order of 120 real pageable pages ... this reduced the time spent in page replacement selection (under relatively heavy load) from approx. 10 percent of total processor to effectively unmeasureable (and at the same time drastically improvement the quality of the replacement choice). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: System 360 EBCDIC vs. ASCII
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well. [EMAIL PROTECTED] (Timothy Sipples) writes: An awful lot of modems and serial connections had to handle 7-bit, too, complicating the user experience for dial-up access to host systems, BBSes, etc. Basically if you set your modem to 7 bits, you struggled to transfer binary files (see: Kermit), and PC extensions for things like line drawing characters looked like a jumbled mess. If you set your modem to 8 bits you usually lost the parity bit, so you lost what little error checking you had. And a lot of systems still tried to use that high order bit for parity, so you saw a jumbled mess on your PC again. Owners of modem dial-up pools installed workarounds to try to detect what the end user had set, but this was a mess, too. On some systems you wouldn't see anything, so you didn't know what to do. (The correct answer: hit Enter a few times, or maybe Escape, or) I'm sure ATT enjoyed some extra earnings as dial-up modem users had to call over and over again, hoping to get the configuration settings right through trial and error, all because of the complications of 7 versus 8 bits. This affected all sorts of serial connections, including hardwired ones: plotters, ASCII terminals, etc. when cp67 was installed at the univ the last week of jan68, it had terminal support for 1052s and 2741s ... but the univ. had some number of tty/ascii devices. so one of the modifications to cp67 was to add tty/ascii terminal support. the base cp67 code had some stuff for dynamically determining the terminal type and switching the 2702 line scanner using the SAD command. so to remain consistent, i worked out a process to add TTY/ascii terminal support ... preserving the base cp67 dynamic terminal type determination. the univ. also was getting dial-up interface ... with base number that would roll-over to the first unused line. the idea that all terminals could dial in on the same phone number, regardless of type. this almost worked ... but it turned out that they had taken some short cuts with 2702 implementation. the issue was that while SAD command would switch the line scanner ... but the short-cut was that the line-speed oscillator was hard-wired to each port. for hard-wired lines ... the appropriate terminal types was connected to the appropriate 2702 with the corresponding line-speed wired (and then cp67 could dynamically determine the correct terminal type and switch the line scanner as needed with the SAD command). However, this wouldn't work for dial-up lines with common dial-in pool ... where any terminal type might get connected to any 2702 port. so somewhat because of this, the univ. decided to build our own clone controller that would also be able to perform dynamic line-speed determination. this involved reverse engineering the 360/67 multiplexor channel interface and building a channel interface board for an Interdata/3 minicomputer (platform for implemented controller clone). misc. past posts about the clone controller project http://www.garlic.com/~lynn/subtopic.html#360pcm i remember two bugs from the project. one bug involved red-lighting the 360/67. the 360/67 had high-resolution timer that tic'ed at approx 13mseconds. the timer had to update loc. 80 storage when it tic'ed. If the timer tic'ed a 2nd time before the previous tic had been updated in storage (say because some channel/controller had obtained the storage bus for the period and failed to release it for that perioid), the timer would force a red-light/machine check. the other bug was initially getting ascii data into storage .. after running it thru standard ascii-ebcdic translation table, it was all garbage. we eventually figured out every byte was bit-reversed ... i.e. 2702 line-scanner would take leading bit off the line and store it in low-order bit position (in a byte ... reversing the order of bits off the line. the interdata/3 started out doing standard ascii taking leading bit off the line and storing it in the high-order bit in a byte. so initially, the ascii bytes was getting to 360/67 main memory in non-bit-reversed bytes and then being run through the standard 2702 ascii-ebcdic (bit-reversed) translation table. this project got written up as the four of us being instrumental in starting the clone controller business. of course, all the clone controller business was the major motivation for the future system project ... lots of past posts http://www.garlic.com/~lynn/subtopic.html#futuresys including a few with this reference http://www.ecole.org/Crisis_and_change_1995_1.htm from above: IBM tried to react by launching a major project called the 'Future System' (FS) in the early 1970's. The idea was to get so far ahead that the competition would never be able to keep up, and to have such a high level of integration that it would be impossible for competitors to follow
Re: CSA 'above the bar'
The following message is a courtesy copy of an article that has been posted to alt.folklore.computers as well. [EMAIL PROTECTED] (Van Dalsen, Herbie) writes: Someone wants to create a shared block of memory CSA/not and share it between programs. My understanding is that a 24-bit program can address 24-bit addresses, 31-bit, 64-bit... So in my inexperienced mind the 24bit program could never share in the happiness of this above the bar heaven of shared storage. as i mentioned in this post http://www.garlic.com/~lynn/2007r.html#62 CSA 'above the bar' ... the way that i originally did sharing implementation and mmap support http://www.garlic.com/~lynn/subtopic.html#mmap was that the same shared object wasn't required to occupy the same virtual address in every virtual address space. however, it could represent a challenge when program images with relocatable address constants were involved http://www.garlic.com/~lynn/subtopic.html#adcon there would still be an issue of the amount of happiness (available in 24bit mode) as opposed to any happiness. it would create a problem for processors that had virtual caches ... i.e. cache lines indexed by virtual address ... resulting in synonyms/duplicates in the cache when the same object was addressed by different virtual addresses. here is old email discussing dual index 3090 D-cache http://www.garlic.com/~lynn/2003j.html#email831118 in this post http://www.garlic.com/~lynn/2003j.html#42 Flash 10208 other posts about virtual cache http://www.garlic.com/~lynn/2006u.html#37 To RISC or not to RISC http://www.garlic.com/~lynn/2006v.html#6 Reasons for the big paradigm switch http://www.garlic.com/~lynn/2006w.html#17 Cache, TLB, and OS one of the other issues for TLB (hardware that translates virtual page addresses to real page addresses) ... all the entries were tagged/associated with specific virtual address spaces ... i.e. STO-associative. This generalized mechanism resulted in a huge number of duplicated entries CSA/common-segment. So as a special case optimization for the whole MVS CSA/common-segment hack gorp ... a special option was provided that identified virtual addresses as something belonging to common-segment. These areas then became associated in the TLB with effectively a system-wide, unique, artificial common-segment virtual address space (effectively violating the whole generalized virtual address space architecture ... rather than associated with generalized virtual address space ... it became associated with a custom operating system specific construct that was known to have very specific characteristics). past post in this thread discussing rise of the whole ugly common segment gorp http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' other posts in this thread http://www.garlic.com/~lynn/2007r.html#64 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#65 CSA 'above the bar' -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: High order bit in 31/24 bit address
The following message is a courtesy copy of an article that has been posted to alt.folklore.computers as well. [EMAIL PROTECTED] (Steve Samson) writes: As for 32-bit mode (TSS) I don't have a POPS for that architecture but I suspect the HO bit is treated as any other. TSS did not use the sign bit as a signal, just as an address bit. lots of 360 documents at bitsavers: http://bitsavers.org/pdf/ibm/360/ including various functional characteristics http://bitsavers.org/pdf/ibm/360/funcChar/ specifically 360/67 functional characteristics a27-2719-0 http://bitsavers.org/pdf/ibm/360/funcChar/A27-2719-0_360-67_funcChar.pdf and ga27-2719-2 http://bitsavers.org/pdf/ibm/360/funcChar/GA27-2719-2_360-67_funcChar.pdf which has a lot of the gory details. as somewhat referenced here ... 360/67 was originally intended for use by tss/360 ... but for a whole variety of reasons, most of them ran cp67 (or in straight 360/65 mode with mvt w/o using virtual memory hardware) http://www.garlic.com/~lynn/2007r.html#64 CSA 'above the bar' curtesy of science center http://www.garlic.com/~lynn/subtopic.html#545tech in any case, psw format, pg. 15 bit meaning 0-3 spare (must be 0) 4 24-32 bit address mode 5 translation control 6 i/o mask (summary) 7 external mask (summary) 8-11protection key 12 ascii-8 mode 13 machine check mask 14 wait state 15 problem state 16-17 instruction length code 18-19 condition code 20-23 program mask 24-31 spare 32-63 instruction address ... there were a quite a few of the machines used internally. one of the projects were adding 370 virtual machine option to cp67 simulation ... this was having cp67 simulate the new instructions added to 370 (prior to announcement of 370 virtual memory). one of the places that deployed numerous of these machines was in the field/data processing/sales division for a project called HONE http://www.garlic.com/~lynn/subtopic.html#hone for hands-on network environment ... the idea was that in the wake of 23jun69 unbundling announcement http://www.garlic.com/~lynn/subtopic.html#unbundle that SEs in the branch office could get operating system hands-on experience with (370) systems running in cp67 (370) virtual machines. however, the science center had also ported apl\360 to cms for cms\apl and done a lot of work enhancing it to operate in large virtual memory environment (most apl\360 was limited to 16k workspaces, hardly adequate for many real world problems). With cms\apl, there were lots of new (internal) apl-based applications developed (some number of them of the genre that today would be done with spreadsheets) ... including configurators ... which basically filled out mainframe system orders for the branch office personal. As the use of these applications grew on HONE ... eventually they eclipsed the virtual guest hands-on training and would consume all available resources. at some point in the 70s, it was not even possible to submit a mainframe order that hadn't been run thru HONE configurator. science center had also done quite a bit of work in the area of sophisticated system performance modeling ... including laying the groundwork for what would become capacity planning. some of this i've commented about with regard to calibrating and validating http://www.garlic.com/~lynn/subtopic.html#benchmark the release of my resource manager http://www.garlic.com/~lynn/subtopic.html#fairshare in addition, a flavor of the performance modeling work was also deployed on HONE as the (apl based) performance predictor. Branch office people could submit customer configuration and workload details/characteristics and then ask what-if questions of the performance predictor ... as to what would happen if there was configuration and/or workload changes. another project was doing the cp67 changes to support a full 370 virtual memory implementation. this had a version cp67 running either in a 360/67 virtual machine (under cp67) or stand-alone real 360/67 simulating virtual machine with full 370 virtual memory operation. Then there was a custom version of cp67 that believed it ran on 370 virtual memory hardware (rather than on 360/67 hardware). This was in regular production use a year before the first engineering 370 machine with virtual memory support was operational (and long before announcement). past posts in the related thread: http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#62 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#64 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#65 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#67 CSA 'above the bar' misc. past posts mentioning performance predictor http://www.garlic.com/~lynn/2001i.html#46 Withdrawal Announcement 901-218 - No More 'small machines' http://www.garlic.com/~lynn/2002b.html#64 ... the
Re: CSA 'above the bar'
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (McKown, John) writes: Just as a thought. Could somebody write a subsystem which starts at IPL time, does the shared GETMAIN, then (here's the rub) somehow have that memory automatically added to every address space which starts thereafter? I don't know enough about subsystems. I would guess that it would be easier for said subsystem to implement a PC so that a client could request access to the shared GCSA (to coin a phrase for it - G for Grande, like the HLASM instructions). The PC would set up all the difficult parts and return a 64-bit address to the shared memory space. re: http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' i had done something similar, but different in the waning days of cp67 and then ported it to vm370. it was generalized memmap function that allowed different virtual address spaces to have the same shared memory object at different addresses. vm370 started out with a drastic subset of this function that was cribbed off the virtual IPL command. however, it was dependent on providing r/o sharing of the same object by segment protection feature that was part of the original, base 370 virtual memory architecture. this was one of the features that got dropped when the retrofit of virtual memory hardware to 370/165 ran into scheduling problems ... could regain six month in schedule if several features were dropped (and the favorite son operating system in pok claimed that they didn't find the features really useful). as a result, this caused all the other processors that already had implemented full 370 virtual memory architecture to go back and pull the dropped features. it also forced the vm370 group to significantly redo their implementation on how to protect shared segments across multiple different virtual address spaces (effectively a real cludge that had been used in cp67) in any case, a drastic subset of my (genealized) memory mapping and sharing implementation was eventually released as something called discontiguous shared segments. lots of past posts mentioning the cms filesystem changes supporting memory mapping (and page mapped operation) http://www.garlic.com/~lynn/subtopic.html#mmap and numerous posts discussing the difficulty that the os/360 relocatable adcon convention represented for allowing sharing same object in different virtual address spaces at potentially different virtual addresses http://www.garlic.com/~lynn/subtopic.html#adcon while tss/360 had numerous other problems, they at least adopted a different convention to address relocatable address constant issue for a shared, virtual memory environment -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: CSA 'above the bar'
[EMAIL PROTECTED] (Binyamin Dissen) writes: Does z/VM use virtual storage? comment in this thread asking how many times has virtual memory been reinvented http://www.garlic.com/~lynn/2007r.html#51 Translation of IBM Basic Assembler to C? some footnotes about the science center http://www.garlic.com/~lynn/subtopic.html#545tech from Melinda's paper VM and the VM Community: Past, Present, and Future http://www.princeton.edu/~melinda ... What was most significant was that the commitment to virtual memory was backed with no successful experience. A system of that period that had implemented virtual memory was the Ferranti Atlas computer, and that was known not to be working well. What was frightening is that nobody who was setting this virtual memory direction at IBM knew why Atlas didn't work ... snip ... quoted from L.W. Comeau, CP-40, the Origin of VM/370, Proceedings of SEAS AM82, September, 1982 and ... Creasy had decided to build CP-40 while riding on the MTA. I launched the effort between Xmas 1964 and year's end, after making the decision while on an MTA bus from Arlington to Cambridge. It was a Tuesday, I believe. (R.J. Creasy, private communication, 1989.) ... snip ... cp40 was built on specially modified 360/40 with virtual memory hardware ... implementing virtual machines. This morphed into cp67 when 360/67 with standard virtual memory became available. and as per previous post in thread http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#62 CSA 'above the bar' the initial hack to mvt for os/vs2, in support of 370 virtual memory, involved borrowing a lot of code from cp67. lots of the vm370 microcode assists developed during the 70s and early 80s eventually morphed into pr/sm and current day LPARs ... which is basically stripped down version of full VM virtual machine function. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: CSA 'above the bar'
[EMAIL PROTECTED] (Ted MacNEIL) writes: That's why there can be a 'double paging' penalty for a LINUX (or z/OS, or z/VM, or...). z/VM, and its predecessors, has always had the capability to defines more storage than is on the box. It even has swap files. i had other problems with the os/vs2 group (initially svs before it morphed into mvs). one was all the stuff about LRU replacement algorithms and what it met. lots of posts on the subject http://www.garlic.com/~lynn/subtopic.html#wsclock early on, the pok performance modeling group had discovered on a page fault that if it selected non-changed pages (for replacement) before changed pages ... there wouldn't need the overhead of doing a write before the read. i tried to convince them it would be violated fundamental tenents of LRU replacement paradigm. It wasn't until well into MVS releases that somebody pointed out that they were selecting for replacement, high-use, non-changed, system/shared executable pages, before (lower use) private application data pages (which were changed/modified). another issue isn't just the double paging overhead ... there is the possibility that a virtual guest is running a LRU-like replacement algorithm and selecting a real page with a low use virtual page for replacement (to be refreshed with the missing page). VM may also be doing LRU-like replacement algorithm and noticed (also) that the guest's real page (virtual machine virtual page) hadn't been recently used and selected it for replacement. The pathelogical problem is that the guest may always be deciding it needs one of its real pages (because the corresponding virtual pages weren't being used) moments after VM has decided to remove the corresponding guest virtual machine page from real storage aka running a virtual guest's LRU-like replacement algorithm can violate the premise behind LRU replacement ... since the guest's real page that corresponds to the guest's least recently used virtual page has some probability of being the next page that the guest might actually decide to use misc. past posts in thread: http://www.garlic.com/~lynn/2007r.html#56 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#62 CSA 'above the bar' http://www.garlic.com/~lynn/2007r.html#64 CSA 'above the bar' -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: CSA 'above the bar'
Steve Samson [EMAIL PROTECTED] writes: The discussion suggests that the dead zone represented an arbitrary decision. However it is absolutely necessary to preserve compatibility with programs dating back to OS/360. If a 24-bit or 31-bit address is interpreted as or expanded to a 64-bit address and the high-order bit happens to be on, that would cast the virtual address into the 2-4 gigabyte range and unpredictable effects could ensue. Use of the high-order bit in an address to signal the end of a parameter list is common, and no practical means of filtering or converting the programs is available. I think the dead zone is necessary in z/VSE for the same reason. Other operating systems did not use the high order bit in the same way, so there is no need for the dead zone in virtual addresses. Has this helped to achieve clarity? 360/67 had both 24-bit and 32-bit virtual addressing modes ... as well as some other things that didn't reappear until xa. there was some discussion in the xa mode about returning to the 360/67 32-bit mode vis-a-vis using 31-bit ... which would have been in the architecture redbook (the discussion i remember was the difference in operation of things like BXH and BXLE instructions between 31-bit and 32-bit modes) principles of operation was one of the first major publications done with cms script ... in large part because it supported conditional so on the command line ... either the whole architecture redbook could be printed ... or just the principles of operation subset (w/o all the additional detail ... it was called redbook because it was distributed in a 3-ring red binder). common segment area started out being the MVS solution to moving subsystems into the own address space ... and the pervasive use of pointer passing APIs. this was what initially led to MVS kernel image occupying 8mbytes of every 16mbyte virtual address space (so for applications making kernel calls ... the kernel could directly access the parameter list). however, this pointer-passing api paradigm created significant problems when subsystems were moved into their own address space (as part of morphing os/vs2 svs to os/vs2 mvs). common segment could start out as 1mbyte in every address space ... where applications could squirrel away parameter list ... and then make call to the subsystem (passing thru the kernel for the address space switch). the problem was for the larger installations, common segment could grow to 5-6 mbytes that appeared in every application virtual address space (with the 8mbyte taken out for the kernel image) that might leave only 2-3mbytes for applications (out of the 16mbytes). the stop-gap solution in the 3033 time-frame was dual-address space mode (pending access registers, program call, etc) ... there was still a pass thru the kernel to switch to a called subsystem ... but the called subsystem could reach back into the calling application's virtual address space (w/o being forced to resorting to the common segment hack). 3033 also introduced a different above the line concept. the mismatch between processor thruput and disk thruput was becoming more and more exacerbated. i once advocated a statement that over a period of a decade or so, that the disk relative system thruput had declined by an order of magnitude (or more) ... aka disk thruput increased by 3-4 times while processor thruput increased by 40-50 times. As a result, real storage was more and more being used for caching and/or other mechanisms to compensate for the lagging disk relative system thruput. we were starting to see clusters of 4341 decked out w/max. storage and max channel and i/o capacity ... matching or beating 3033 thruput at a lower price. one of the 4341 cluster benefits was that there was more aggregate real storage than the 16mbyte limit for 3033. the hack was to redefine two (undefined/unused) bits in the page table entry. standard page table entry had 16 bits, including a 12bit (4k) page number field (allowed addressing up to 16mbytes real storage). With the two additional bits, it was possible to address up to 16384 4kbyte pages (up to 64mbyte of real storage) ... but only 16mbytes at a time. in real addressing mode ... it was only possible to address the first 16mbytes and in virtual addressing mode ... it was only possible to address a specific 16mbytes (but it was possible to have more than 4096 total 4kbyte pages, some of which could reside about 16mbyte real). it was possible to use channel program IDAL to specify address greater than 16mbyte real address (allowing data to be read/written above the 16mbyte line). however, the actual channel programs were still limited to residing below the 16mbyte line. some of this was masked by the whole channel program translation mechanism that was necessary as part of moving to virtual memory environment. the original transition for mvt was hacking a little bit of support for a single virtual address space (i.e. os/vs2 svs) and
Re: IBM System/3 3277-1
The following message is a courtesy copy of an article that has been posted to comp.sys.ibm.sys3x.misc,alt.folklore.computers,bit.listserv.ibm-main as well. Rostyslaw J. Lewyckyj [EMAIL PROTECTED] writes: If memory hasn't failed me, we read mark sense cards on something that was called a 1230. We didn't have one in the computing center. It was in a separate laboratory somewhere in the School of Education. We sent the decks over there. I don't remember what we got back. I think the 1230 may have punched the marked card. re: http://www.garlic.com/~lynn/2007q.html#71 IBM System/3 3277-1 http://www.garlic.com/~lynn/2007r.html#2 IBM System/3 3277-1 wiki mark sense page http://en.wikipedia.org/wiki/Mark_sense mentions that 513, 514, 557, and 519 could handle mark sense. also has pointer to 805 test scoring machine. 513 514 reproducing punches could handle mark sense ... so it is possible that a 513/514 had preprossed the mark sense student registration cards ... and the 2540 was only processing the reproduced punch cards (and i just not paying that much attention). the wiki reference also has url for 513/514 (pdf) reference manual -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: IBM System/3 3277-1
The following message is a courtesy copy of an article that has been posted to comp.sys.ibm.sys3x.misc,alt.folklore.computers,bit.listserv.ibm-main as well. Anne Lynn Wheeler [EMAIL PROTECTED] writes: 3277 had quite a bit of local intelligence ... it was possible to do some custom stuff in the terminal that changed the repeat start-delay and repeat ... as well as adding fifo to handle keyboard locking up if you happen to be typing when the system went to (re)write something on the screen. the move to 3274 controller for 3278/3279/etc terminals ... moved all that intelligence back into the controller ... reducing amount of electronics and manufacturing costs. with electronics moved back into controller ... it also degraded performance and response. re: http://www.garlic.com/~lynn/2007r.html#7 IBM System/3 3277-1 http://www.garlic.com/~lynn/2007r.html#8 IBM System/3 3277-1 somebody picking around in some of the referenced old postings, sent private email asking about reference to ANR download being 2-3 times baster than DCA download ... and what was ANR ... other than APPN Automatic Networking Routing. ANR was3272/3277 ... vis-a-vis DCA 3274/3278-9. In addition to DCA having slower human (real terminal) response ... because so much of the electronics had been moved back into controller, it also affected later terminal emulation download thruput. quicky search engine for 3277 anr turns up http://www.classiccmp.org/pipermail/cctech/2007-September/084640.html misc. past posts mentioning terminal emulation http://www.garlic.com/~lynn/subnetwork.html#emulation as client/server started to proliferate ... the communication group made various attempts (like SAA) to protect their terminal emulation install base. when we came up with 3tier/multi-tier architecture ... we took lots of heat from the sna and saa forces. misc. posts mentioning coming up with multitier networking architecture http://www.garlic.com/~lynn/subnetwork.html#3tier for other drift ... APPN started out as AWP164. For a time, the person responsible and I used to report to the same executive. I would periodically chide him that the communication group didn't appreciate what he was doing and that he should instead work on real networking (like tcp/ip). In fact, the communication group non-concurred with announcing APPN. After some delay and escalation, the announcement letter was carefully rewritten to not state any connection between APPN and SNA. of course we were also running hsdt project ... misc. posts http://www.garlic.com/~lynn/subnetwork.html#hsdt and recent post illustrating gap between what we were doing and what the communication group was doing http://www.garlic.com/~lynn/2007p.html#64 part of the issue was that in early days of SNA ... my wife had co-authored AWP39 ... peer-to-peer networking ... which the communication group possibly viewed as competitive with their communication activity. she was then con'ed into going to pok to be in charge of loosely-coupled architecture and was frequently battling with SNA forces that it wasn't appropriate for loosely-coupled operation. She came up with peer-coupled shared data architecture ... which didn't see a lot of uptake until sysplex ... except for IMS hot-standby ... misc. past references http://www.garlic.com/~lynn/subtopic.html#shareddata recent posts mentioning ASWP39 http://www.garlic.com/~lynn/2007b.html#9 Mainframe vs. Server (Was Just another example of mainframe http://www.garlic.com/~lynn/2007b.html#48 6400 impact printer http://www.garlic.com/~lynn/2007d.html#55 Is computer history taugh now? http://www.garlic.com/~lynn/2007h.html#35 sizeof() was: The Perfect Computer - 36 bits? http://www.garlic.com/~lynn/2007h.html#39 sizeof() was: The Perfect Computer - 36 bits? http://www.garlic.com/~lynn/2007l.html#62 Friday musings on the future of 3270 applications http://www.garlic.com/~lynn/2007o.html#72 FICON tape drive? http://www.garlic.com/~lynn/2007p.html#12 JES2 or JES3, Which one is older? http://www.garlic.com/~lynn/2007p.html#23 Newsweek article--baby boomers and computers http://www.garlic.com/~lynn/2007q.html#46 Are there tasks that don't play by WLM's rules -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: IBM System/3 3277-1
The following message is a courtesy copy of an article that has been posted to comp.sys.ibm.sys3x.misc,alt.folklore.computers,bit.listserv.ibm-main as well. bbreynolds [EMAIL PROTECTED] writes: This thread started about the 3277-001 used on a System/3 Model 15 (would that be a 5415?): as 3277's relied on the 3271/3272/3275 for the major portion of their intelligence, I would assume that there would have had to been some pretty substantial hardware in the System/3 to make the 3277-001 believe it was attached to a controller. I can't think how the functions would be split out on a 3277 not on a controller; unless the 3277-001 was gutted. Any hint if a cable other than a simple coax connected the 3277 to the CPU? 3277 had quite a bit of local intelligence ... it was possible to do some custom stuff in the terminal that changed the repeat start-delay and repeat ... as well as adding fifo to handle keyboard locking up if you happen to be typing when the system went to (re)write something on the screen. the move to 3274 controller for 3278/3279/etc terminals ... moved all that intelligence back into the controller ... reducing amount of electronics and manufacturing costs. with electronics moved back into controller ... it also degraded performance and response. several of us complained about it ... but were told that 327x terminals were targeted at data entry market and didn't have the requirements for interactive response and human factors that would be needed for something like interactive computing. as seen in some of the referenced performance comparisons ... say http://www.garlic.com/~lynn/2001m.html#19 3270 protocol ... it was much more difficult to achieve subsecond response with 3274/3278 vis-a-vis 3272/3277. However, for mvs/tso with system response already on the order of a second (or much worse) ... it was pretty negligible consideration. however, heavily loaded vm/cms systems tended to be more on the order of a quarter second (or less, one system i had carefeeding of ... was on the order of .11 seconds 90th percentile for trivial interactive under heavy load). past posts mentioning some (hardware) fixes to 3277 ... and not being able to doing anything with later 3278/3279 because even that bit of electronics had been moved back into the controller (and/or some other 3272/3277 issues vis-a-vis 3274/3278). http://www.garlic.com/~lynn/94.html#23 CP spooling programming technology http://www.garlic.com/~lynn/98.html#49 Edsger Dijkstra: the blackest week of his professional life http://www.garlic.com/~lynn/99.html#28 IBM S/360 http://www.garlic.com/~lynn/99.html#69 System/1 ? http://www.garlic.com/~lynn/99.html#193 Back to the original mainframe model? http://www.garlic.com/~lynn/99.html#239 IBM UC info http://www.garlic.com/~lynn/2000c.html#63 Does the word mainframe still have a meaning? http://www.garlic.com/~lynn/2000c.html#65 Does the word mainframe still have a meaning? http://www.garlic.com/~lynn/2000c.html#66 Does the word mainframe still have a meaning? http://www.garlic.com/~lynn/2000c.html#67 Does the word mainframe still have a meaning? http://www.garlic.com/~lynn/2000d.html#12 4341 was Is a VAX a mainframe? http://www.garlic.com/~lynn/2000g.html#23 IBM's mess http://www.garlic.com/~lynn/2001b.html#12 Now early Arpanet security http://www.garlic.com/~lynn/2001f.html#49 any 70's era supercomputers that ran as slow as today's supercompu http://www.garlic.com/~lynn/2001i.html#51 DARPA was: Short Watson Biography http://www.garlic.com/~lynn/2001k.html#30 3270 protocol http://www.garlic.com/~lynn/2001k.html#33 3270 protocol http://www.garlic.com/~lynn/2001k.html#44 3270 protocol http://www.garlic.com/~lynn/2001k.html#46 3270 protocol http://www.garlic.com/~lynn/2001l.html#32 mainframe question http://www.garlic.com/~lynn/2001m.html#17 3270 protocol http://www.garlic.com/~lynn/2001m.html#19 3270 protocol http://www.garlic.com/~lynn/2002f.html#14 Mail system scalability (Was: Re: Itanium troubles) http://www.garlic.com/~lynn/2002i.html#43 CDC6600 - just how powerful a machine was it? http://www.garlic.com/~lynn/2002i.html#48 CDC6600 - just how powerful a machine was it? http://www.garlic.com/~lynn/2002i.html#50 CDC6600 - just how powerful a machine was it? http://www.garlic.com/~lynn/2002j.html#67 Total Computing Power http://www.garlic.com/~lynn/2002j.html#74 Itanium2 power limited? http://www.garlic.com/~lynn/2002j.html#77 IBM 327x terminals and controllers (was Re: Itanium2 power http://www.garlic.com/~lynn/2002k.html#2 IBM 327x terminals and controllers (was Re: Itanium2 power http://www.garlic.com/~lynn/2002k.html#6 IBM 327x terminals and controllers (was Re: Itanium2 power http://www.garlic.com/~lynn/2002m.html#24 Original K R C Compilers http://www.garlic.com/~lynn/2002p.html#29 Vector display systems http://www.garlic.com/~lynn/2002q.html#51 windows office xp http://www.garlic.com/~lynn/2003b.html#29 360/370 disk drives