beust@taloa.unice.fr (Cedric Beust) (06/02/91)
I wanted to have statistics about the CPU use on my Amiga and was
very suprised to see that gomf (3.0, so launched via runback) used
about 12 (twelve!) percent of CPU time, and in READY mode, not
WAITING. So I decided not to use it any more.
Anyone can corroborate this amazing figure?
+------------------------------------------------------------------------+
| Cedric BEUST University of Nice |
| INET: beust@taloa.unice.fr $whoami |
| UUCP: llaor.unice.fr!arkonis!beust god (personal alias) |
| -- "To be, or not to be...", |
| That is illogical, captain! |
| -- Spock |
+------------------------------------------------------------------------+u8705377@cs.uow.edu.au (Paul Anthony Wilkinson) (06/03/91)
beust@taloa.unice.fr (Cedric Beust) writes: > I wanted to have statistics about the CPU use on my Amiga and was >very suprised to see that gomf (3.0, so launched via runback) used >about 12 (twelve!) percent of CPU time, and in READY mode, not >WAITING. So I decided not to use it any more. > Anyone can corroborate this amazing figure? I stopped using Gomf at version 2.0 when I found that it was using a large chunk of CPU time. I suspect it is because it frequently checks lowmem to see if someone stomped on it via a NULL pointer etc. I also found that when GOMF did catch my programs that left locks open & stuff and I had to reboot anyway. The simpler solution is just not to make programming errors 8-) . Regards, Paul Wilkinson u8705377@wraith.cs.uow.edu.au - I can't afford a .sig let alone opinions.
griffin@frith.egr.msu.edu (Danny Griffin) (06/03/91)
beust@taloa.unice.fr (Cedric Beust) writes: > I wanted to have statistics about the CPU use on my Amiga and was >very suprised to see that gomf (3.0, so launched via runback) used >about 12 (twelve!) percent of CPU time, and in READY mode, not >WAITING. So I decided not to use it any more. Is that with 10 other programs running or by itself? -- Dan Griffin griffin@frith.egr.msu.edu
GELSON%SBU.UFRGS.ANRS.BR@uicvm.uic.edu (Gelson Dias Santos) (06/03/91)
Paul Anthony Wilkinson <u8705377@cs.uow.edu.au> writes: >beust@taloa.unice.fr (Cedric Beust) writes: >> I wanted to have statistics about the CPU use on my Amiga and was >>very suprised to see that gomf (3.0, so launched via runback) used >>about 12 (twelve!) percent of CPU time, and in READY mode, not >>WAITING. So I decided not to use it any more. >I stopped using Gomf at version 2.0 when I found that it was using a large >chunk of CPU time. I suspect it is because it frequently checks lowmem to see >if someone stomped on it via a NULL pointer etc. I also found that when >GOMF did catch my programs that left locks open & stuff and I had to reboot >anyway. The simpler solution is just not to make programming errors 8-) . Two questions: Is there any PD util to show CPU statistics, or are you using a commercial product? What is Gomf, anyway? ;-) * Gelson Dias Santos * Bitnet/Internet: GELSON@SBU.UFRGS.ANRS.BR * * Porto Alegre - RS * Sorry, english is not my native language! * * BRAZIL * (Alguem fala portugues?) *
bombadil@diku.dk (Kristian Nielsen) (06/04/91)
beust@taloa.unice.fr (Cedric Beust) writes: > I wanted to have statistics about the CPU use on my Amiga and was >very suprised to see that gomf (3.0, so launched via runback) used >about 12 (twelve!) percent of CPU time, and in READY mode, not >WAITING. So I decided not to use it any more. How did you obtain this figure? If you used Xoper (which is reasonable, since it is in many ways an excelent and very useful program), note that it doesn't count cpu time, but instead the number of task switches. But more important, if you didn't use Xoper, what did you use? Is it PD / Freeware / Shareware? I really miss an accurate means of timing the cpu-usage of my programs. - Kristian
mrimages@beach.gal.utexas.edu (06/04/91)
XOper has a function that displays percent CPU used, and it says that it itself uses only 3-5% of active CPU time. It also has a myriad of other functions that are quite helpful. R. Luebbert INET: mrimages@beach.gal.utexas.edu BNET: mrimages@utmbeach
specter@disk.uucp (Byron Max Guernsey) (06/04/91)
GELSON%SBU.UFRGS.ANRS.BR@uicvm.uic.edu (Gelson Dias Santos) writes: >Two questions: Is there any PD util to show CPU statistics, or are you using a >commercial product? > What is Gomf, anyway? ;-) there is a shareware program called XOPER that displays all kinds of statistics about tasks, ports, memory, etc. GOMF stands for Get Outa My Face and it catches guru meditation errors before they crash you totally and allows you to safely reboot. (don't think it would let you survive a guru) >* Gelson Dias Santos * Bitnet/Internet: GELSON@SBU.UFRGS.ANRS.BR * >* Porto Alegre - RS * Sorry, english is not my native language! * >* BRAZIL * (Alguem fala portugues?) * Byron
alex@bilver.uucp (Alex Matulich) (06/04/91)
In article <1991Jun3.111119.8148@cs.uow.edu.au> u8705377@cs.uow.edu.au (Paul Anthony Wilkinson) writes: >beust@taloa.unice.fr (Cedric Beust) writes: >> I wanted to have statistics about the CPU use on my Amiga and was >>very suprised to see that gomf (3.0, so launched via runback) used >>about 12 (twelve!) percent of CPU time, and in READY mode, not >>WAITING. So I decided not to use it any more. My understanding is that GOMF is one of those tasks that dynamically adjusts its CPU use. If your Amiga is idling, GOMF will use more. Anyone know differently? >I stopped using Gomf at version 2.0 when I found that it was using a large >chunk of CPU time. I suspect it is because it frequently checks lowmem to see No, it doesn't. You need to use another utility like MemGuard for this. >if someone stomped on it via a NULL pointer etc. I also found that when >GOMF did catch my programs that left locks open & stuff and I had to reboot >anyway. The simpler solution is just not to make programming errors 8-) . GOMF has saved my ass so many times while programming, that I do not want to give it up. True, I often re-boot anyway after a crash, but I like the added comfort of being able to clean up complicated projects, examine temporary ramdisk files, etc, before re-booting. In actual timing tests, I have not noticed GOMF to slow down any application that I run. -- _ |__ Alex Matulich /(+__> Unicorn Research Corp, 4621 N Landmark Dr, Orlando, FL 32817 //| \ UUCP: alex@bilver.uucp <or> ...uunet!tarpit!bilver!alex ///__) bitnet: IN%"bilver!alex@uunet.uu.net"
griffin@frith.egr.msu.edu (Danny Griffin) (06/05/91)
alex@bilver.uucp (Alex Matulich) writes: >GOMF has saved my ass so many times while programming, that I do not want >to give it up. True, I often re-boot anyway after a crash, but I like the >added comfort of being able to clean up complicated projects, examine >temporary ramdisk files, etc, before re-booting. >In actual timing tests, I have not noticed GOMF to slow down any >application that I run. I'd like to echo the above statements. Also, since I am usually doing 18 things at once (whether or not they're actually being used - usually 2 or 3 are big apps, other are small background things) the Amiga may occasionally hiccup if i get too anxious with dmouse flipping between them. Even on those errors that become unrecoverable, at least *that* app is in suspension while i can go to all others and save my previous two hours of PageStream layouts, emacs editing, finish a download, etc. Try using a program like AIBB (Amiga Intuition Based Benchmarks) and check the speed of your Amiga with and without your background processes one by one. -- Dan Griffin griffin@frith.egr.msu.edu
beust@taloa.unice.fr (Cedric Beust) (06/07/91)
In article <1991Jun3.171023.13315@odin.diku.dk>, bombadil@diku.dk (Kristian Nielsen) writes: : : How did you obtain this figure? If you used Xoper (which is reasonable, : since it is in many ways an excelent and very useful program), note that it : doesn't count cpu time, but instead the number of task switches. But more : important, if you didn't use Xoper, what did you use? Is it PD / Freeware / : Shareware? I really miss an accurate means of timing the cpu-usage of my : programs. I used a ps written by a friend of mine. I wanted to ask him how he counted the percentage but didn't have time to do so. I didn't think of using XOper, though. I'll try this ASAP. If people are interested in this ps, I'll upload it to ab20. Is there an ultimate way of evaluating CPU time on the Amiga? +------------------------------------------------------------------------+ | Cedric BEUST University of Nice | | INET: beust@taloa.unice.fr $whoami | | UUCP: llaor.unice.fr!arkonis!beust god (personal alias) | | -- "To be, or not to be...", | | That is illogical, captain! | | -- Spock | +------------------------------------------------------------------------+
stevex@artech.UUCP (Steve Tibbett) (06/08/91)
In article <223@taloa.unice.fr> beust@taloa.unice.fr (Cedric Beust) writes: > > I wanted to have statistics about the CPU use on my Amiga and was >very suprised to see that gomf (3.0, so launched via runback) used >about 12 (twelve!) percent of CPU time, and in READY mode, not >WAITING. So I decided not to use it any more. > > Anyone can corroborate this amazing figure? And I bet you used XOPER or some other thing that measures the dispatched counter that Exec keeps to figure out how busy the task is. I finally figured out why VirusX and my other programs are getting such an apparently high number there: If a program does a loop like this: while (TRUE) { WaitTOF(); if (This==That) { Do Stuff; }; }; Then it's going to look like it's taking a very high amount of CPU time, because it's getting dispatched 60 times a second - but it's not actually slowing the machine down that much, or using much CPU time at all (near zero). I bet this is what GOMF is doing. I know it's what my programs are doing, and I don't see anything wrong with it - Complain to whoever wrote your CPU time measurer. -- ...Steve's Signature (when Steve's at work)...
beust@taloa.unice.fr (Cedric Beust) (06/14/91)
In article <stevex.3270@artech.UUCP>, stevex@artech.UUCP (Steve Tibbett) writes: : : And I bet you used XOPER or some other thing that measures the dispatched : counter that Exec keeps to figure out how busy the task is. : : <deleted> : : I bet this is what GOMF is doing. I know it's what my programs are doing, : and I don't see anything wrong with it - Complain to whoever wrote your CPU : time measurer. Right. I didn't use Xoper but since then I've seen the figures vary considerably from one measurer to the other. And anyway, there is no ultimate way to measure the CPU time on the Amiga :-( I'll keep GOMF3 in a drawer and will take it out when necessary. BTW, Steve, do you plan to release a new version of VirusX? +------------------------------------------------------------------------+ | Cedric BEUST University of Nice | | INET: beust@taloa.unice.fr $whoami | | UUCP: llaor.unice.fr!arkonis!beust god (personal alias) | | -- "To be, or not to be...", | | That is illogical, captain! | | -- Spock | +------------------------------------------------------------------------+
stevex@artech.UUCP (Steve Tibbett) (06/18/91)
In article <240@taloa.unice.fr> beust@taloa.unice.fr (Cedric Beust) writes: >In article <stevex.3270@artech.UUCP>, stevex@artech.UUCP (Steve Tibbett) writes: >: >: And I bet you used XOPER or some other thing that measures the dispatched >: counter that Exec keeps to figure out how busy the task is. >: >: I bet this is what GOMF is doing. I know it's what my programs are doing, >: and I don't see anything wrong with it - Complain to whoever wrote your CPU >: time measurer. > > Right. I didn't use Xoper but since then I've seen the figures vary > considerably from one measurer to the other. And anyway, there is > no ultimate way to measure the CPU time on the Amiga :-( Sure there is - start an interrupt and sample the ExecBase->ThisTask. I wonder what ThisTask points to when no task is currently "Running"? > I'll keep GOMF3 in a drawer and will take it out when necessary. > > BTW, Steve, do you plan to release a new version of VirusX? When I finish BBX, I'll get back into VirusX and do a new version, probably a rewrite, and by that time, probably 2.0-specific. -- ...Steve's Signature (when Steve's at work)...
thomas@diku.dk (Thomas Nikolajsen) (06/21/91)
stevex@artech.UUCP (Steve Tibbett) writes: >> Right. I didn't use Xoper but since then I've seen the figures vary >> considerably from one measurer to the other. And anyway, there is >> no ultimate way to measure the CPU time on the Amiga :-( >Sure there is - start an interrupt and sample the ExecBase->ThisTask. I wonder >what ThisTask points to when no task is currently "Running"? Not a good idea, where did you se that method documented? Try using tc_Switch and tc_Launch (all tasks), whats between is idle time, same method to measure time used by tasks. >> I'll keep GOMF3 in a drawer and will take it out when necessary. I tried the above, and GOMF (version 2.xx) is not that CPU hungry (circa 5%). > ...Steve's Signature (when Steve's at work)... I tried to reach Steve with email, but it bounced at artech (stevex unknown). thomas