felix@ai.sri.com (Francois Felix INGRAND) (08/14/90)
I have notice that GNUEmacs is using "lot of CPU" on my Sparc 1+. Here is a top 10, and as you can see, although I am doing nothing, the 2 gnuemacs are using 22% each. (The X server is getting a good share too). All these process are "idle". Does that mean that emacs is continuously checking for inputs and therefore, forcing the system in heavy context switching activities? Is is possible to avoid this problem? last pid: 1550; load averages: 1.40, 1.36, 1.02 16:44:55 51 processes: 49 sleeping, 2 running Cpu states: 18.4% user, 0.0% nice, 81.6% system, 0.0% idle Memory: 35920K (14608K) real, 16648K free PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 1481 felix 49 0 2536K 2648K run 2:27 29.34% 29.30% Xsun 1503 felix 1 0 4448K 3296K sleep 2:01 22.30% 22.27% gnuemacs 1493 felix 1 0 3352K 2176K sleep 1:45 21.12% 21.09% gnuemacs 1531 felix 25 0 352K 752K run 0:06 0.00% 0.00% top 1463 felix 15 0 568K 840K sleep 0:00 0.00% 0.00% tcsh 1526 felix 15 0 544K 816K sleep 0:00 0.00% 0.00% tcsh 1511 felix 15 0 104K 344K sleep 0:00 0.00% 0.00% loadst 1510 felix 15 0 104K 344K sleep 0:00 0.00% 0.00% loadst 152 root 15 0 24K 16K sleep 0:35 0.00% 0.00% update 1480 felix 5 0 88K 584K sleep 0:00 0.00% 0.00% xinit As shown on the vmstat, cs is quite high... cayucos% felix [22]:vmstat 1 procs memory page disk faults cpu r b w avm fre re at pi po fr de sr d0 d1 d2 s3 in sy cs us sy id 1 0 0 0 8492 0189 7 4 7 0 1 1 0 0 0 591690 296 11 29 60 1 0 0 0 8492 0 6 0 0 0 0 0 0 0 0 0 765157 869 17 83 0 2 0 0 0 8480 0 4 0 0 0 0 0 0 0 0 0 855230 879 18 82 0 2 0 0 0 8480 0 3 0 0 0 0 0 0 0 0 0 1145212 877 17 83 0 2 0 0 0 8480 0 2 0 0 0 0 0 0 0 0 0 1395215 878 22 78 0 2 0 0 0 8480 0 1 0 0 0 0 0 0 0 0 0 1435264 885 22 78 0 2 0 0 0 8480 0 0 0 0 0 0 0 0 0 0 0 1225354 899 22 78 0 2 0 0 0 8436 0 0 0 0 0 0 0 0 0 0 0 1005442 913 23 77 0 2 0 0 0 8436 0 2 0 0 0 0 0 3 0 0 0 865453 916 16 84 0 -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Francois Felix INGRAND SRI International, AIC felix@AI.SRI.COM 333, Ravenswood Avenue (415) 859-5584 MENLO PARK, CA 94025, USA "Read my Lisp... No new syntax" (nil)
meissner@osf.org (Michael Meissner) (08/14/90)
In article <FELIX.90Aug13165019@cayucos.ai.sri.com> felix@ai.sri.com (Francois Felix INGRAND) writes: | I have notice that GNUEmacs is using "lot of CPU" on my Sparc 1+. | | Here is a top 10, and as you can see, although I am doing nothing, the 2 | gnuemacs are using 22% each. (The X server is getting a good share too). | | All these process are "idle". Does that mean that emacs is continuously | checking for inputs and therefore, forcing the system in heavy context | switching activities? | | Is is possible to avoid this problem? I don't know if it helps, but I used to see this on a Sun3 if the X server died, or emacs was left running after a session ended. By making sure emacs was done before shutting down a session I avoided the problem. -- Michael Meissner email: meissner@osf.org phone: 617-621-8861 Open Software Foundation, 11 Cambridge Center, Cambridge, MA, 02142 Do apple growers tell their kids money doesn't grow on bushes?
felix@ai.sri.com (Francois Felix INGRAND) (08/15/90)
In article <FELIX.90Aug13165019@cayucos.ai.sri.com> felix@ai.sri.com (Francois Felix INGRAND) writes: > >From: felix@ai.sri.com (Francois Felix INGRAND) >Newsgroups: comp.emacs,gnu.emacs >Date: 13 Aug 90 23:50:19 GMT > >I have notice that GNUEmacs is using "lot of CPU" on my Sparc 1+. > >Here is a top 10, and as you can see, although I am doing nothing, the 2 >gnuemacs are using 22% each. (The X server is getting a good share too). > >All these process are "idle". Does that mean that emacs is continuously >checking for inputs and therefore, forcing the system in heavy context >switching activities? It looks like this problem only occurs when you run gnuemacs under X and most likely on Sparcstation . I cannot reproduce it under SUNVIEW. tmb@ai.mit.edu (Thomas M. Breuel) wrote: " GNU Emacs will sometimes get into a mode where it keeps talking to the X server. This is really harmless and doesn't seem to slow down the machine significantly. " My feeling is that nevertheless, it is eating some CPU on my sparc. All these context switch reported by vmstat are not "free", and are probably responsible for the high system time (80%). So I am not so sure that this is harmless. I would appreciate some comments from GNUEmacs/X/Sparc gurus on this matter. last pid: 1550; load averages: 1.40, 1.36, 1.02 16:44:55 51 processes: 49 sleeping, 2 running Cpu states: 18.4% user, 0.0% nice, 81.6% system, 0.0% idle Memory: 35920K (14608K) real, 16648K free PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 1481 felix 49 0 2536K 2648K run 2:27 29.34% 29.30% Xsun 1503 felix 1 0 4448K 3296K sleep 2:01 22.30% 22.27% gnuemacs 1493 felix 1 0 3352K 2176K sleep 1:45 21.12% 21.09% gnuemacs 1531 felix 25 0 352K 752K run 0:06 0.00% 0.00% top 1463 felix 15 0 568K 840K sleep 0:00 0.00% 0.00% tcsh 1526 felix 15 0 544K 816K sleep 0:00 0.00% 0.00% tcsh 1511 felix 15 0 104K 344K sleep 0:00 0.00% 0.00% loadst 1510 felix 15 0 104K 344K sleep 0:00 0.00% 0.00% loadst 152 root 15 0 24K 16K sleep 0:35 0.00% 0.00% update 1480 felix 5 0 88K 584K sleep 0:00 0.00% 0.00% xinit As shown on the vmstat, cs is quite high... cayucos% felix [22]:vmstat 1 procs memory page disk faults cpu r b w avm fre re at pi po fr de sr d0 d1 d2 s3 in sy cs us sy id 1 0 0 0 8492 0189 7 4 7 0 1 1 0 0 0 591690 296 11 29 60 1 0 0 0 8492 0 6 0 0 0 0 0 0 0 0 0 765157 869 17 83 0 2 0 0 0 8480 0 4 0 0 0 0 0 0 0 0 0 855230 879 18 82 0 2 0 0 0 8480 0 3 0 0 0 0 0 0 0 0 0 1145212 877 17 83 0 2 0 0 0 8480 0 2 0 0 0 0 0 0 0 0 0 1395215 878 22 78 0 2 0 0 0 8480 0 1 0 0 0 0 0 0 0 0 0 1435264 885 22 78 0 2 0 0 0 8480 0 0 0 0 0 0 0 0 0 0 0 1225354 899 22 78 0 2 0 0 0 8436 0 0 0 0 0 0 0 0 0 0 0 1005442 913 23 77 0 2 0 0 0 8436 0 2 0 0 0 0 0 3 0 0 0 865453 916 16 84 0 -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Francois Felix INGRAND SRI International, AIC felix@AI.SRI.COM 333, Ravenswood Avenue (415) 859-5584 MENLO PARK, CA 94025, USA "Read my Lisp... No new syntax" (nil)
cs@Eng.Sun.COM (Carl Smith) (08/15/90)
In article <FELIX.90Aug13165019@cayucos.ai.sri.com>, felix@ai.sri.com (Francois Felix INGRAND) writes: > > I have notice that GNUEmacs is using "lot of CPU" on my Sparc 1+. ... > Is is possible to avoid this problem? I stumbled onto a workaround trying to investigate this. Simply trace each of the offending processes (e.g. ``trace -p 1503'') for a bit. You'll notice rather a lot of output at first, then it will become quiescent. Carl
duty@ariel.its.unimelb.edu.au (Duty Programmer) (08/15/90)
In article <FELIX.90Aug13165019@cayucos.ai.sri.com> felix@ai.sri.com (Francois Felix INGRAND) writes: > >I have notice that GNUEmacs is using "lot of CPU" on my Sparc 1+. I have noticed the same thing running MicroEmacs under AOS/VS ! Now, I don't have the source code handy, but does Emacs do function calls for every keystroke? If it does, then this could slow things up a bit...? rab -- Richard Brown | E-mail: rab@ariel.its.unimelb.EDU.AU School of Physics | Phone : +61 3 344 5081 University of Melbourne | Fax : +61 3 347 4783 Parkville Victoria AUSTRALIA 3052 | Telex : AA35185
shore@mtxinu.COM (Melinda Shore) (08/16/90)
In article <204@ariel.its.unimelb.edu.au> duty@ariel.its.unimelb.edu.au (Duty Programmer) writes: >Now, I don't >have the source code handy, but does Emacs do function calls for every >keystroke? Function calls? Heck, I did some tracing under Mach and found that emacs did 20-30 *system calls* per keystroke under X. >If it does, then this could slow things up a bit...? A tad. -- Melinda Shore shore@mtxinu.com mt Xinu ..!uunet!mtxinu.com!shore