Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Sun, 24 May 2009 20:22:37 +0200 (CEST) Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote, *again* without attribution: From the glossary (p. 630) of _The_Design_and_Implementation_of_the _FreeBSD_Operating_System_ by McKusick and Neville-Neil: load average A measure of CPU load on the system. The load average in FreeBSD is an average of the number of processes ready to run or waiting for short-term events such as disk I/O to complete, as sampled once per second over the previous one- minute interval of system operation. so this glossary should be fixed because it's nonsense. first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. You are mistaken. I think what you are referring to is the percentage of CPU usage, not CPU load. Note that CPU usage can never exceed 100% per CPU, laying aside confusion over application of this term to hyperthreading chips, whereas CPU load obviously can be much higher. It might help you to think of CPU load as CPU commitment. If you wish to design an operating system and then write its documentation, you will, of course, be free to define terms relating to your system in whatever manner you like. However, the authors of the book are not only among the architects of FreeBSD, but also 4.?BSD UNIX and possibly earlier versions of BSD UNIX. They have been good enough to document what they have done and to define what they mean by terms used in that documentation and throughout the system, which most of the non-self-righteously arrogant members of the community appreciate. Treat the information in this book as having come from the horses' mouths, excepting only those parts that have been added/ changed/deleted since FreeBSD 5.2. FWIW, the concept of load average in UNIX is quite old. I don't know exactly how old it is, but it may date back to 7th Ed. UNIX from ATT or even earlier. BSD UNIX branched off about that time or possibly 6th Ed. The definition of load average cited above is essentially the average length of the queue of runnable processes (including those that are currently running) plus the processes expected to be runnable in the immediate future, e.g., much sooner than a hypothetical process that has been swapped out could be made runnable. Scott Bennett, Comm. ASMELG, CFIAG ** * Internet: bennett at cs.niu.edu * ** * A well regulated and disciplined militia, is at all times a good * * objection to the introduction of that bane of all free governments * * -- a standing army. * *-- Gov. John Hancock, New York Journal, 28 January 1790 * ** ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Yuri wrote: Look below: load over 7 and no processes take much CPU. Yuri 7.2-PRERELEASE, 32-bit on i7-920. last pid: 93192; load averages: 7.68, 6.27, 4.61 up 2+03:11:29 20:25:24 204 processes: 9 running, 193 sleeping, 1 stopped, 1 zombie CPU: 5.3% user, 0.0% nice, 0.0% system, 0.0% interrupt, 94.7% idle Mem: 867M Active, 1684M Inact, 279M Wired, 65M Cache, 112M Buf, 92M Free Swap: 16G Total, 142M Used, 16G Free Couple of possible reasons: 1) You have many short lived processes that are spawned, do some work and die (some kind of web server?). You can't see them because they live too shortly. See if the last pid is rapidly increasing. Also, hit H to display individual threads. 2) You have kernel processes that are doing some intermittent work. Hit S to see them. In any case, see the line where it says 9 running processes? This is where the load average comes from. The CPU utilization is not directly related to the load average. Load average is not scaled to NCPU - a count of 7 (or in your case, aymptotically 9) means there are 7 (or 9) processes wanting to run. The global CPU utilization (the CPU: line) *is* scaled to NCPU - 100% here means all CPUs are busy all the time. Individual processes' CPU utilization *isn't* scaled to NCPU. A process taking 100% CPU on its own means it only requires / runs on 1 CPU. A multithreading process can have, for example, 400% CPU utilization and the global CPU utilization can be = 100%. signature.asc Description: OpenPGP digital signature
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
2009/5/24 Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: From the glossary (p. 630) of _The_Design_and_Implementation_of_the _FreeBSD_Operating_System_ by McKusick and Neville-Neil: load average A measure of CPU load on the system. The load average in FreeBSD is an average of the number of processes ready to run or waiting for short-term events such as disk I/O to complete, as sampled once per second over the previous one- minute interval of system operation. so this glossary should be fixed because it's nonsense. first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. Er, what? Of course it is! Chris -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. You are mistaken. I think what you are referring to is the percentage of no i'm not. doing lots of I/O and little CPU load produces high load average. the explanation from the book is wrong. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. Er, what? Of course it is! amount of disk I/O is a measure of CPU load? seems you are true expert ;) ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
2009/5/25 Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. Er, what? Of course it is! amount of disk I/O is a measure of CPU load? seems you are true expert ;) Do you ever think before you type? You regularly fill this mailing list with crap, incorrect advice, and correcting experts on topics that you haven't got a clue on. Just google load average and see for yourself. Remember checking things before making oneself look a fool? Perhaps you used to do that at one time, most other people do. Chris -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Chris Rees wrote: 2009/5/25 Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. Er, what? Of course it is! amount of disk I/O is a measure of CPU load? seems you are true expert ;) Do you ever think before you type? You regularly fill this mailing list with crap, incorrect advice, and correcting experts on topics that you haven't got a clue on. Just google load average and see for yourself. Remember checking things before making oneself look a fool? Perhaps you used to do that at one time, most other people do. Chris I think Wojciech means '...which is NOT measure of CPU _utilization_' In that case he's correct: whenever the CPU has to wait a lot for I/O, like network and disk, then the _load_ will go up, while the CPU _utilization_ stays low. Peter -- http://www.boosten.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
2009/5/25 Peter Boosten pe...@boosten.org: Chris Rees wrote: 2009/5/25 Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. Er, what? Of course it is! amount of disk I/O is a measure of CPU load? seems you are true expert ;) Do you ever think before you type? You regularly fill this mailing list with crap, incorrect advice, and correcting experts on topics that you haven't got a clue on. Just google load average and see for yourself. Remember checking things before making oneself look a fool? Perhaps you used to do that at one time, most other people do. Chris I think Wojciech means '...which is NOT measure of CPU _utilization_' In that case he's correct: whenever the CPU has to wait a lot for I/O, like network and disk, then the _load_ will go up, while the CPU _utilization_ stays low. Peter I appreciate that while English may not be people's first language, and most have excellent skills (including Wojciech, don't get me wrong) there is still a responsibility to make sure you can be understood. He has been trying to mock me in my understanding of what he wrote: amount of disk I/O is a measure of CPU load? seems you are true expert ;) If you're going to do this, you should make sure that what you WROTE, not mean, is correct. Otherwise, misinformation spreads and is saved in these archives for someone to stumble upon. Chris -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Do you ever think before you type? You regularly fill this mailing list with crap please don't name things crap just because you don't understand ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
I think Wojciech means '...which is NOT measure of CPU _utilization_' exactly what i said. In that case he's correct: whenever the CPU has to wait a lot for I/O, like network and disk, then the _load_ will go up, while the CPU _utilization_ stays low. and that's inconsistent with explanation that load average is measure of CPU load. it's not. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On 25 mei 2009, at 19:12, Wojciech Puchar wrote: I think Wojciech means '...which is NOT measure of CPU _utilization_' exactly what i said. Regardless from what you said: you _wrote_ CPU _load_, not cpu _utilization_, which are two completely different thingemies. The load averages in top describe the state the entire machine is in, not just the CPU. In that case he's correct: whenever the CPU has to wait a lot for I/ O, like network and disk, then the _load_ will go up, while the CPU _utilization_ stays low. and that's inconsistent with explanation that load average is measure of CPU load. it's not. I never claimed load average = CPU load! Peter -- http://www.boosten.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Regardless from what you said: you _wrote_ CPU _load_, not cpu what's a difference for you between CPU load and CPU utilization? i mean CPU load not system load. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On 25 mei 2009, at 21:08, Wojciech Puchar wrote: Regardless from what you said: you _wrote_ CPU _load_, not cpu what's a difference for you between CPU load and CPU utilization? i mean CPU load not system load. The CPU will perform the same, whether at 10% utilization, or at 100% utilization, the system however won't. That's the difference between load and utilization. Peter -- http://www.boosten.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
The CPU will perform the same, whether at 10% utilization, or at 100% utilization, the system however won't. That's the difference between load and utilization. still don't understand you. CPU will not perform the same at 10% utilization, it will perform 10 times less than at 100% utilization. CPU load == CPU utilization == how big percentage of time CPU (or CPUs by average) are doing anything except being in idle loop or hlt/waiting for interrupt. it's exactly the same words in that context. load average is how much processes (by average) is not doing calculations because something is not yet available and depends of computer resources(*) - like CPU time, disk I/O results etc.. (*) - for example waiting on tty read is not calculated to load average as it's depends on human not computer. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Mon, May 25, 2009 at 3:24 PM, Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote: The CPU will perform the same, whether at 10% utilization, or at 100% utilization, the system however won't. That's the difference between load and utilization. still don't understand you. CPU will not perform the same at 10% utilization, it will perform 10 times less than at 100% utilization. You are assuming this based on what? CPU load == CPU utilization == how big percentage of time CPU (or CPUs by average) are doing anything except being in idle loop or hlt/waiting for interrupt. it's exactly the same words in that context. Not true. top(1) can fully utilize the CPU. Doing so does not put the system under full load. load average is how much processes (by average) is not doing calculations because something is not yet available and depends of computer resources(*) - like CPU time, disk I/O results etc.. (*) - for example waiting on tty read is not calculated to load average as it's depends on human not computer. What? -- Glen Barber ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On 25 mei 2009, at 21:24, Wojciech Puchar wrote: The CPU will perform the same, whether at 10% utilization, or at 100% utilization, the system however won't. That's the difference between load and utilization. still don't understand you. CPU will not perform the same at 10% utilization, it will perform 10 times less than at 100% utilization. *sigh* The CPU = Central Processing Unit will perform it's calculations at so many megahertz while at 10% utilization or at 100% utilization. The entire machine is only performing at 10% of its capacity, in your statement above. Load = burden. Under heavy _load_ the _machine_ will become sluggish, but the CPU will still be performing at the same megahertz speed. You cannot put the CPU under load (=burden), since it was designed to perform at 100% (actually it can be, if the queue length gets too large, and then it's called load, but that's not being done in top). The fact that the CPU has to wait for some I/O will not influence the performance of the CPU, but to the entire machine. CPU load == CPU utilization == how big percentage of time CPU (or CPUs by average) are doing anything except being in idle loop or hlt/ waiting for interrupt. it's exactly the same words in that context. load average is how much processes (by average) is not doing calculations because something is not yet available and depends of computer resources(*) - like CPU time, disk I/O results etc.. Actually, it's a combination of both running and waiting processes. (*) - for example waiting on tty read is not calculated to load average as it's depends on human not computer. Peter -- http://www.boosten.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Not true. top(1) can fully utilize the CPU. Doing so does not put the system under full load. top uses small percentage of CPU power. if it would use all - it WOULD mean full CPU load. load average is how much processes (by average) is not doing calculations because something is not yet available and depends of computer resources(*) - like CPU time, disk I/O results etc.. (*) - for example waiting on tty read is not calculated to load average as it's depends on human not computer. What? exactly what i wrote. reread if you don't understand. anyway i'm quitting this discussion as it's obvious for most people that can read what i mean and what CPU load/utilization mean. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
The CPU = Central Processing Unit will perform it's calculations at so many megahertz while at 10% utilization or at 100% utilization. The entire machine no. it will not. all today x86 CPUs reacts on HLT command and doesn't do anything except waiting for interrupt (and saving lots of power). is only performing at 10% of its capacity, in your statement above. no - CPU is performing at 10% of it's capacity. entire machine (disks, network etc.) is another thing. Actually, it's a combination of both running and waiting processes. yes that's true. i missed this - all waiting for machine resource to be available+those which are presently calculating something. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On 25 mei 2009, at 21:37, Wojciech Puchar wrote: The CPU = Central Processing Unit will perform it's calculations at so many megahertz while at 10% utilization or at 100% utilization. The entire machine no. it will not. all today x86 CPUs reacts on HLT command and doesn't do anything except waiting for interrupt (and saving lots of power). Do you actually read back what you write: you're saying here that when a CPU has only 10% utilization, it'll run slower than when performing at 100%... I'm giving up ;-) Peter -- http://www.boosten.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
2009/5/25 Wojciech Puchar woj...@wojtek.tensor.gdynia.pl: you are funny. Don't ever send me private messages like that. You are a troll who gives harmful and misleading advice. Chris -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Mon, 25 May 2009 21:42:40 +0200 Peter Boosten pe...@boosten.org wrote: On 25 mei 2009, at 21:37, Wojciech Puchar wrote: The CPU = Central Processing Unit will perform it's calculations at so many megahertz while at 10% utilization or at 100% utilization. The entire machine no. it will not. all today x86 CPUs reacts on HLT command and doesn't do anything except waiting for interrupt (and saving lots of power). Do you actually read back what you write: you're saying here that when a CPU has only 10% utilization, it'll run slower than when performing at 100%... While it's not the default behaviour, if you run powerd(8) then yes, the CPU will run slower when it's less utilized. My laptop will run at between 150MHz and 2500MHz depending on the current workload. -- Bruce Cran ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
you are funny. Don't ever send me private messages like that. You are a troll who gives harmful and misleading advice. exactly because of sentences like that i think you are funny. sorry - but you moved this to public ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Do you actually read back what you write: you're saying here that when a CPU has only 10% utilization, it'll run slower than when performing at 100%... i said it perform 10 times less work than when 100% utilized. exactly - read back again. I'm giving up ;-) looks like you just want to prove that i'm wrong, nothing else. so give up. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
performing at 100%... While it's not the default behaviour, if you run powerd(8) then yes, the CPU will run slower when it's less utilized. that's extra,and very useful option. anyway even without that modern processor gets MUCH less power just when being halted by hlt instruction. there is still some power used for clock network within chip, and gate oxide leakage, but almost no other gate switching. just using hlt takes down power usage to 1/3-1/5 of full power. reducing clock with turn it down even more. i'm not sure but turning modern processors on lower speed this way (powerd) change voltage down too a bit which even reduces leakage. anyway powerd didn't work well on my laptop last time i tried, but it was with FreeBSD 6, maybe something changed - i will try this. maybe it will add some 10-20 minutes more battery life. anyway lots of power is used by my UMTS modem, and i can't disconnect it as i usually do need the net while using laptop. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Yuri wrote: Look below: load over 7 and no processes take much CPU. Yuri 7.2-PRERELEASE, 32-bit on i7-920. last pid: 93192; load averages: 7.68, 6.27, 4.61 up 2+03:11:29 20:25:24 204 processes: 9 running, 193 sleeping, 1 stopped, 1 zombie CPU: 5.3% user, 0.0% nice, 0.0% system, 0.0% interrupt, 94.7% idle Mem: 867M Active, 1684M Inact, 279M Wired, 65M Cache, 112M Buf, 92M Free Swap: 16G Total, 142M Used, 16G Free PID USERNAMETHR PRI NICE SIZERES STATE C TIME WCPU COMMAND 60032 yuri 1 460 285M 183M select 0 41:15 0.59% Xorg 60400 yuri 1 40 12576K 9144K kqread 4 29:44 0.00% wineserver 92982 yuri 1 440 53012K 16800K CPU3 3 18:50 0.00% kdeinit4 92986 yuri 1 440 53012K 16800K CPU7 7 18:48 0.00% kdeinit4 92988 yuri 1 1070 53012K 16840K CPU6 6 17:22 0.00% kdeinit4 60104 yuri 1 440 132M 45860K select 0 16:58 0.00% kwin 92984 yuri 1 1170 53012K 16800K RUN5 14:56 0.00% kdeinit4 60096 yuri 1 440 89732K 30040K select 4 10:10 0.00% kded4 93141 yuri 1 530 53012K 16800K CPU5 5 3:52 0.00% kdeinit4 93139 yuri 1 440 53012K 16800K CPU1 1 3:30 0.00% kdeinit4 60174 yuri 1 440 3168K 1400K select 0 1:28 0.00% ksysguardd 450 root 1 40 3128K 800K select 4 0:44 0.00% dhclient 1131 messagebus1 40 3344K 1384K select 4 0:40 0.00% dbus-daemon Sure. This is not an uncommon occurrence really. The load average is the number of processes in the queue for a CPU time slice averaged over 5, 10 or 15 minutes. For multi-core systems the LA is scaled by the number of cores so a LA of 1.0 means all cores have active processes pretty much continually. Now, you might think that an active process will take the CPU utilisation to 100%, but that is not necessarily so. Some numerical applications can do that, but purely CPU bound processes are relatively uncommon in everyday usage. In actuality what happens is that the processor will need to retrieve data from somewhere to operate on. There's a hierarchy of data stores of various speeds (latency, rather than bandwidth): L1 Cache L2 Cache L3 Cache Main RAM Disk Network Where the L1 Cache is accessible in a few clock ticks (nanoseconds), Main RAM can take microseconds to access, disk can take milliseconds to access, and Network can take 10 -- 1000s of milliseconds. Or in other words, about 9 orders of magnitude difference. So when the data you need to process is too big to fit in the fastest caches, or when it comes from a particularly slow location or when you have a lot of active processes causing context switches, then the CPU core will be making frequent IO requests and spending time waiting for them to be fulfilled. Now, for sources like disks and network where the retrieval is much slower than the typical timescale of events on the CPU the process will yield the CPU to something else and only get a new timeslice once the IO request has been fulfilled. For an access to main RAM however that form of yielding is less likely. Consequently the CPU can end up waiting for 100s of clock cycles until it gets some bytes to process. In the mean time, other processes are also sitting in the queue wanting CPU time slices -- hence the high LA with low CPU utilization. Scheduling CPU timeslices to make maximum use of available resources is the difference between a really performant OS and a disaster. A good scheduler is the critical central piece of code around which the rest of an OS can be constructed. Combine that with the complexity of having multiple cores, and that threads of execution sometimes have to be moved to different cores, and on other occasions sometimes need to stick to the same core in order to make best use of resources and you will start to appreciate quite how hard it is to write a good scheduler. Unsurprisingly, the design of such things is a matter of fairly impassioned debate amongst the rarified circle of people capable of writing them. That sort of argument was the genesis of the FreeBSD / DragonflyBSD fork a few years back. You can rest assured though that FreeBSD certainly does have one of the very best schedulers currently available and it is specifically targeted at getting the best out of the sort of multicore CPUs available nowadays. Cheers, Matthew -- Dr Matthew J Seaman MA, D.Phil. 7 Priory Courtyard Flat 3 PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate Kent, CT11 9PW signature.asc Description: OpenPGP digital signature
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Hi, Matthew On Sun, May 24, 2009 at 3:46 AM, Matthew Seaman m.sea...@infracaninophile.co.uk wrote: Yuri wrote: [snip] Sure. This is not an uncommon occurrence really. The load average is the number of processes in the queue for a CPU time slice averaged over 5, 10 or 15 minutes. For multi-core systems the LA is scaled by the number of cores so a LA of 1.0 means all cores have active processes pretty much continually. I thought, if it was a dual-core for example, a load average of 1.00 would indicate 50% CPU utilization overall (1 process using only 1 core)[1]. 2.00 on a dual-core would be 100%, 3.00 on a dual-core would be 100% utilization, and always 1 process in the wait queue, and so on. Now, you might think that an active process will take the CPU utilisation to 100%, but that is not necessarily so. Some numerical applications can do that, but purely CPU bound processes are relatively uncommon in everyday usage. In actuality what happens is that the processor will need to retrieve data from somewhere to operate on. There's a hierarchy of data stores of various speeds (latency, rather than bandwidth): L1 Cache L2 Cache L3 Cache Main RAM Disk Network Does this affect the load average though? My understanding was that if the CPU cannot immediately process data, the data gets put into the wait queue until L2 Cache (then RAM, etc, etc) returns the data to be processed. [1] - http://www.teamquest.com/resources/gunther/display/5/ (not necessarily a reputable source I suppose, but explains it well...) -- Glen Barber 570.328.0318 ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Glen Barber wrote: Hi, Matthew On Sun, May 24, 2009 at 3:46 AM, Matthew Seaman m.sea...@infracaninophile.co.uk wrote: Yuri wrote: [snip] Sure. This is not an uncommon occurrence really. The load average is the number of processes in the queue for a CPU time slice averaged over 5, 10 or 15 minutes. For multi-core systems the LA is scaled by the number of cores so a LA of 1.0 means all cores have active processes pretty much continually. I thought, if it was a dual-core for example, a load average of 1.00 would indicate 50% CPU utilization overall (1 process using only 1 core)[1]. 2.00 on a dual-core would be 100%, 3.00 on a dual-core would be 100% utilization, and always 1 process in the wait queue, and so on. It seems both ways have been used in different OSes, which is confusing. A quick test of a single threaded process that will spin one CPU on a multi-core FreeBSD box shows the value is /not/ scaled by the number of cores. Which means that the LA the OP was talking about is actually a lot less alarming than it originally appears. It's clear from the top output that his machine has at least 8 cores, so a LA of 7 is really not very heavily loaded. Now, you might think that an active process will take the CPU utilisation to 100%, but that is not necessarily so. Some numerical applications can do that, but purely CPU bound processes are relatively uncommon in everyday usage. In actuality what happens is that the processor will need to retrieve data from somewhere to operate on. There's a hierarchy of data stores of various speeds (latency, rather than bandwidth): L1 Cache L2 Cache L3 Cache Main RAM Disk Network Does this affect the load average though? My understanding was that if the CPU cannot immediately process data, the data gets put into the wait queue until L2 Cache (then RAM, etc, etc) returns the data to be processed. Yes it does: when a process is on the CPU and blocked waiting for IO it does not necessarily yield the CPU to another process. It depends on timescales -- obviously if the CPU will have to wait milliseconds for data it makes no sense to block other processes. Waiting a few microseconds is a different matter though: it might take that long to load up L2/L3 cache with that processes' working data, so yielding the CPU for that sort of delay would mean the process never got run, which is counter productive... It helps if the working set is already in the L3 cache -- so having the correct amount[*] of cache RAM available is an important design criterion. It's something that Intel was shown to have got wrong with some of the Pentium series chips when a low powered Pentium M designed for mobile use smoked a much higher clock speed Pentium chip designed for all-out server use simply because it had about 4x as much cache. Cheers, Matthew [*] ie. as much as possible. -- Dr Matthew J Seaman MA, D.Phil. 7 Priory Courtyard Flat 3 PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate Kent, CT11 9PW signature.asc Description: OpenPGP digital signature
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Sun, May 24, 2009 at 5:07 AM, Matthew Seaman wrote: I thought, if it was a dual-core for example, a load average of 1.00 would indicate 50% CPU utilization overall (1 process using only 1 core)[1]. 2.00 on a dual-core would be 100%, 3.00 on a dual-core would be 100% utilization, and always 1 process in the wait queue, and so on. It seems both ways have been used in different OSes, which is confusing. A quick test of a single threaded process that will spin one CPU on a multi-core FreeBSD box shows the value is /not/ scaled by the number of cores. Meaning a load average of 1.00 on a single-core versus dual-core means the same thing? I can't tell if you said what I said (or meant) with different wording, or if you said the opposite. :-) Which means that the LA the OP was talking about is actually a lot less alarming than it originally appears. It's clear from the top output that his machine has at least 8 cores, so a LA of 7 is really not very heavily loaded. So in this situation, he has 1 core idle all of the time, correct? Does this affect the load average though? My understanding was that if the CPU cannot immediately process data, the data gets put into the wait queue until L2 Cache (then RAM, etc, etc) returns the data to be processed. Yes it does: when a process is on the CPU and blocked waiting for IO it does not necessarily yield the CPU to another process. It depends on timescales -- obviously if the CPU will have to wait milliseconds for data it makes no sense to block other processes. Waiting a few microseconds is a different matter though: it might take that long to load up L2/L3 cache with that processes' working data, so yielding the CPU for that sort of delay would mean the process never got run, which is counter productive... It helps if the working set is already in the L3 cache -- so having the correct amount[*] of cache RAM available is an important design criterion. Makes sense. -- Glen Barber ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Look below: load over 7 and no processes take much CPU. load average is NOT sum of CPU loads. for example program reading constantly from HDD and using no CPU will add 1 to load average. other things like net I/O etc. are calculated too. i can't explain you exactly how because i don't know precisely. but load average is total load not just CPU load Yuri 7.2-PRERELEASE, 32-bit on i7-920. last pid: 93192; load averages: 7.68, 6.27, 4.61 up 2+03:11:29 20:25:24 204 processes: 9 running, 193 sleeping, 1 stopped, 1 zombie CPU: 5.3% user, 0.0% nice, 0.0% system, 0.0% interrupt, 94.7% idle Mem: 867M Active, 1684M Inact, 279M Wired, 65M Cache, 112M Buf, 92M Free Swap: 16G Total, 142M Used, 16G Free PID USERNAMETHR PRI NICE SIZERES STATE C TIME WCPU COMMAND 60032 yuri 1 460 285M 183M select 0 41:15 0.59% Xorg 60400 yuri 1 40 12576K 9144K kqread 4 29:44 0.00% wineserver 92982 yuri 1 440 53012K 16800K CPU3 3 18:50 0.00% kdeinit4 92986 yuri 1 440 53012K 16800K CPU7 7 18:48 0.00% kdeinit4 92988 yuri 1 1070 53012K 16840K CPU6 6 17:22 0.00% kdeinit4 60104 yuri 1 440 132M 45860K select 0 16:58 0.00% kwin 92984 yuri 1 1170 53012K 16800K RUN5 14:56 0.00% kdeinit4 60096 yuri 1 440 89732K 30040K select 4 10:10 0.00% kded4 93141 yuri 1 530 53012K 16800K CPU5 5 3:52 0.00% kdeinit4 93139 yuri 1 440 53012K 16800K CPU1 1 3:30 0.00% kdeinit4 60174 yuri 1 440 3168K 1400K select 0 1:28 0.00% ksysguardd 450 root 1 40 3128K 800K select 4 0:44 0.00% dhclient 1131 messagebus1 40 3344K 1384K select 4 0:40 0.00% dbus-daemon ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Sun, 24 May 2009 11:57:08 +0200 (CEST) Wojciech Puchar woj...@wojtek.tensor.gdynia.pl wrote without proper attribution: Look below: load over 7 and no processes take much CPU. load average is NOT sum of CPU loads. for example program reading constantly from HDD and using no CPU will add 1 to load average. other things like net I/O etc. are calculated too. i can't explain you exactly how because i don't know precisely. but load average is total load not just CPU load From the glossary (p. 630) of _The_Design_and_Implementation_of_the _FreeBSD_Operating_System_ by McKusick and Neville-Neil: load average A measure of CPU load on the system. The load average in FreeBSD is an average of the number of processes ready to run or waiting for short-term events such as disk I/O to complete, as sampled once per second over the previous one- minute interval of system operation. In the same volume in the discussion of Calculations of Thread Priority by the 4.4 BSD scheduler (p. 101), it says, ... the *load* is a sampled average of the sum of the lengths of the run queue and of the short-term sleep queue over the previous 1-minute interval of system operation. Seems pretty straightforward to me. Scott Bennett, Comm. ASMELG, CFIAG ** * Internet: bennett at cs.niu.edu * ** * A well regulated and disciplined militia, is at all times a good * * objection to the introduction of that bane of all free governments * * -- a standing army. * *-- Gov. John Hancock, New York Journal, 28 January 1790 * ** ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
From the glossary (p. 630) of _The_Design_and_Implementation_of_the _FreeBSD_Operating_System_ by McKusick and Neville-Neil: load average A measure of CPU load on the system. The load average in FreeBSD is an average of the number of processes ready to run or waiting for short-term events such as disk I/O to complete, as sampled once per second over the previous one- minute interval of system operation. so this glossary should be fixed because it's nonsense. first - says that it's measure of CPU load then - or waiting for short-term events such as disk I/O - which is NOT measure of CPU load. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
This guy advises buying an old G4 Mac laptop to use as a netbook: http://lowendmac.com/ed/herlihy/09ph/ibook-netbook.html While Apple might be planning to stop supporting PowerPC, one could run FreeBSD on it. Mac-Pro has good prices on used Mac laptops. A G4 PowerBook is $500 to $650 depending on what kind of burner is installed. http://www.mac-pro.com/s.nl/sc.2/category.66/.f I was just now looking into ARM netbooks. I think there's only one actual shipping model so far, but ARM shows great promise because ARM CPUs use very little power. I expect there will be lots of them by the end of the year. Is there a FreeBSD ARM port? There's not one for 7.2. Mike -- Michael David Crawford m...@prgmr.com prgmr.com - We Don't Assume You Are Stupid. Xen-Powered Virtual Private Servers: http://prgmr.com/xen ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Sun, 24 May 2009 12:02:41 -0700, Michael David Crawford m...@prgmr.com wrote: Mac-Pro has good prices on used Mac laptops. A G4 PowerBook is $500 to $650 depending on what kind of burner is installed. http://www.mac-pro.com/s.nl/sc.2/category.66/.f Hmmm... I still think about reviving my iBook G4, maybe it gives a good Netbook-lookalike. :-) I was just now looking into ARM netbooks. I think there's only one actual shipping model so far, but ARM shows great promise because ARM CPUs use very little power. I expect there will be lots of them by the end of the year. Thre has been an interesting article on OSNews lately: http://www.osnews.com/story/21530/The_Loongson-2_MIPS_Lemote_Yeeloong_Netbook Maybe this is interesting, too. Is there a FreeBSD ARM port? There's not one for 7.2. I'm not aware of one, but I think NetBSD has it. But finally, NetBSD isn't FreeBSD. :-) -- Polytropon From Magdeburg, Germany Happy FreeBSD user since 4.0 Andra moi ennepe, Mousa, ... ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
I was just now looking into ARM netbooks. I think there's only one actual shipping model so far, but ARM shows great promise because ARM CPUs use very little power. I expect there will be lots of them by the end of the year. Is there a FreeBSD ARM port? There's not one for 7.2. there are for some ARM CPUs in source tree. Mike -- Michael David Crawford m...@prgmr.com prgmr.com - We Don't Assume You Are Stupid. Xen-Powered Virtual Private Servers: http://prgmr.com/xen ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Is there a FreeBSD ARM port? There's not one for 7.2. I'm not aware of one, but I think NetBSD has it. But finally, NetBSD isn't FreeBSD. :-) quite a big difference. was enough for me to switch to FreeBSD some time ago. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
On Sun, May 24, 2009 at 3:02 PM, Michael David Crawford m...@prgmr.com wrote: This guy advises buying an old G4 Mac laptop to use as a netbook: http://lowendmac.com/ed/herlihy/09ph/ibook-netbook.html While Apple might be planning to stop supporting PowerPC, one could run FreeBSD on it. Mac-Pro has good prices on used Mac laptops. A G4 PowerBook is $500 to $650 depending on what kind of burner is installed. http://www.mac-pro.com/s.nl/sc.2/category.66/.f I was just now looking into ARM netbooks. I think there's only one actual shipping model so far, but ARM shows great promise because ARM CPUs use very little power. I expect there will be lots of them by the end of the year. Is there a FreeBSD ARM port? There's not one for 7.2. How did this topic get switched to netbooks? -- Glen Barber ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
How can this 'top' command output make sense? Load over 7 and total CPU use ~5%
Look below: load over 7 and no processes take much CPU. Yuri 7.2-PRERELEASE, 32-bit on i7-920. last pid: 93192; load averages: 7.68, 6.27, 4.61 up 2+03:11:29 20:25:24 204 processes: 9 running, 193 sleeping, 1 stopped, 1 zombie CPU: 5.3% user, 0.0% nice, 0.0% system, 0.0% interrupt, 94.7% idle Mem: 867M Active, 1684M Inact, 279M Wired, 65M Cache, 112M Buf, 92M Free Swap: 16G Total, 142M Used, 16G Free PID USERNAMETHR PRI NICE SIZERES STATE C TIME WCPU COMMAND 60032 yuri 1 460 285M 183M select 0 41:15 0.59% Xorg 60400 yuri 1 40 12576K 9144K kqread 4 29:44 0.00% wineserver 92982 yuri 1 440 53012K 16800K CPU3 3 18:50 0.00% kdeinit4 92986 yuri 1 440 53012K 16800K CPU7 7 18:48 0.00% kdeinit4 92988 yuri 1 1070 53012K 16840K CPU6 6 17:22 0.00% kdeinit4 60104 yuri 1 440 132M 45860K select 0 16:58 0.00% kwin 92984 yuri 1 1170 53012K 16800K RUN5 14:56 0.00% kdeinit4 60096 yuri 1 440 89732K 30040K select 4 10:10 0.00% kded4 93141 yuri 1 530 53012K 16800K CPU5 5 3:52 0.00% kdeinit4 93139 yuri 1 440 53012K 16800K CPU1 1 3:30 0.00% kdeinit4 60174 yuri 1 440 3168K 1400K select 0 1:28 0.00% ksysguardd 450 root 1 40 3128K 800K select 4 0:44 0.00% dhclient 1131 messagebus1 40 3344K 1384K select 4 0:40 0.00% dbus-daemon ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
