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: > there was 370/195 with outboard channels (basically an upgraded of > 360/195 with few additional instructions ... but never got any of the > virtual memory support upgrade). > ... > sjr/bldg.28 was still running an aging MVT 370/195 system when the > bldg.15 disk product test lab got an early engineering 3033. Carefully > pipelined ... 370/195 would peak at about twice 3033 sustained thruput > ... however most codes ran about same thruput on 370/195 as 3033. The re: http://www.garlic.com/~lynn/2009r.html#51 "Portable" data centers 370/195 was pipelined and to get peak sustained thruput ... just about had to be carefully constructed loops totally within the pipeline. 370/195 didn't have branch prediction or speculative execution ... so (a non-carefully constructed pipeline looping branch) would drain the pipeline. most codes had frequent branches keeping the pipeline from being more than half-full and running at half peak thruput (or about the same as 3033). the 370/195 group came up with idea for hyperthreading ... basically emulate two processor smp ... with two PSWs, two sets of registers, etc. ... but no additional pipeline or execution units. the idea was if single execution stream only kept the pipeline half-full at half-thruput ... then two independent execution streams might be able to keep the pipeline full and the execution units fully busy. i got invited to participate because of other work on smp. however, the effort never got as far as even being announced. for other drift ... one of the nails in the "future system" coffin http://www.garlic.com/~lynn/submain.html#futuresys was analysis that if a FS machine was built from 195 technology ... applications running on such a machine would have thruput of 370/145 (that was one of the FS sections that had a complete enuf definition that analysis could be perform ... lots of the other FS was so inadequately defined that couldn't be analyzed). The actual comparison was Eastern Airlines reservation system (System/one), optimized for running on 370/195 ... the comparison pointed out that the best thruput it could expect to achieve (on the highest performance FS system) was that of running on 370/145 another site of some hardware & FS discussion from the FS period: http://www.jfsowa.com/computer/memo125.htm a few misc. tidbits from above: Whereas the Model 168 had required 40 months to evolve from development to initial shipment, the 3033 was shipped to its first customer after only 28 months in development." IBM achieved that feat by remapping the Model 168 design into the circuitry intended for VANDERBILT and making a number of tweaks in the design to improve performance. Of course, IBM could have delivered a machine with similar or better performance in 1975 instead of 1977, if they hadn't killed all the System/370 design projects to avoid competition with the FS fantasy. ... snip ... later the Eastern Airline res system was the basis for Amadeus ... and my wife severed a short stint as chief architect. Unfortunately she sided with the selection of x.25 (over sna) ... and the SNA forces managed to have her replaced. However, it didn't do them any good, since Amadeus went with x.25 anyway ... a couple recent posts mentioning Amadeus http://www.garlic.com/~lynn/2009j.html#33 IBM touts encryption innovation http://www.garlic.com/~lynn/2009l.html#55 IBM halves mainframe Linux engine prices -- 40+yrs virtualization experience (since Jan68), online at home since Mar1970 ---------------------------------------------------------------------- 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
