Messages in this thread | | | Date | Tue, 27 Oct 1998 12:34:57 -0500 | From | Noah Beck <> | Subject | Re: 2.1.126 SMP scheduling |
| |
Well, I don't know the answer to your problem, but to add another data point, my quad PPro with 4 simultaneous CPU-bound processes is also pretty much unusable (in X at the local display). This is certainly not new behavior, though. Currently, 2.1.125 exhibits this, and 2.1.106 certainly did as well (the only two kernels I've tried on here). Running an additional CPU-intensive job does not affect responsiveness, but the additional job does not seem to suffer the same penalty as interactive response suffers.
If, for example, I run four parallel loops as below, the interactive remote performance of the system is something like, if I hold down the <return> key so tcsh prints out a bunch of prompts, I only get about one or two prompts printed per second on average. If one were to move the mouse on the X console during this time, it would be extremely jumpy. Attempting to interactively edit files is an exercise in cursor prediction, as well.
60 noah@goodheads ~/bench % time ./loop2 69.130u 0.000s 1:09.13 100.0% 0+0k 0+0io 111pf+0w 61 noah@goodheads ~/bench % time ./loop2 & ; time ./loop2 & ; time ./loop2 & ; time ./loop2 & [1] 506 [2] 507 [3] 508 [4] 509 62 noah@goodheads ~/bench % [2] Done ./loop2 70.750u 0.010s 1:10.77 99.9% 0+0k 0+0io 111pf+0w 62 noah@goodheads ~/bench % [3] Done ./loop2 71.410u 0.030s 1:11.44 100.0% 0+0k 0+0io 111pf+0w 62 noah@goodheads ~/bench % [4] Done ./loop2 73.700u 0.000s 1:13.72 99.9% 0+0k 0+0io 111pf+0w 62 noah@goodheads ~/bench % [1] Done ./loop2 79.520u 0.020s 1:19.55 99.9% 0+0k 0+0io 111pf+0w 62 noah@goodheads ~/bench % time ./loop2 & ; time ./loop2 & ; time ./loop2 & ; time ./loop2 & ; time ./loop2 & [1] 511 [2] 512 [3] 513 [4] 514 [5] 515 63 noah@goodheads ~/bench % [5] Done ./loop2 74.660u 0.020s 1:19.80 93.5% 0+0k 0+0io 111pf+0w 63 noah@goodheads ~/bench % [4] Done ./loop2 75.380u 0.060s 1:32.68 81.3% 0+0k 0+0io 111pf+0w 63 noah@goodheads ~/bench % [3] Done ./loop2 74.160u 0.000s 1:32.89 79.8% 0+0k 0+0io 111pf+0w 63 noah@goodheads ~/bench % [2] Done ./loop2 74.610u 0.010s 1:34.09 79.3% 0+0k 0+0io 111pf+0w 63 noah@goodheads ~/bench % [1] Done ./loop2 75.930u 0.060s 1:35.30 79.7% 0+0k 0+0io 111pf+0w 63 noah@goodheads ~/bench % cat loop2.C main(){ for(signed long i=0xc0000000; i<0x7fffffff;i++) ; }
That was compiled with no optimizations, btw.
Noah
Scott McNab wrote: > > Hello. > > Has process scheduling been changed significantly between 2.1.125 and > 2.1.126? On my dual PII-300 the responsiveness of the machine when both > processors are loaded seems considerably worse. > > Under 2.1.125 I could run 2 l3enc processes simultaneously with normal > priority and it would not visibly disturb the X server for most > activities. > > However, using 2.1.126 I find that as soon as I start these 2 processes > the mouse cursor under X becomes very jumpy and slow to respond making > the system unpleasant to use. > > I am not sure whether it is due to a change in the scheduling algorithm > or whether there is some other problem but kswapd seems to take a lot > more CPU cycles than previously (according to top). > > Has anyone else noticed similar symptoms? Please CC me in the reply as > I'm not subscribed to this list. > > ------------------------------------------------------------------------ > Scott McNab Phone: +61 8 9386 0540 > Software Engineer Student Support Officer > Research and Development Dept University Computing Services > Fractal Graphics University of Western Australia > http://www.fractal.csiro.au/ http://www.uwa.edu.au/ > sdm@fractal.csiro.au jedi@tartarus.uwa.edu.au > > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.rutgers.edu > Please read the FAQ at http://www.tux.org/lkml/
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |