I have been using and developing a trivial response benchmark which is
intended to give a reproducable measure of system response to trivial
tasks, such as uncovering a window or typing a small command such as
'ls' requiring modest memory and CPU.
The benchmark runs a noload case, then forks a load process (or
processes) in the background, pauses long enough to simulate user
interaction (and get swapped out if the system is memory stressed), then
reports the time it took to complete, including the ratio of loaded to
noload time.
I got some results so bad I thought the program was in error, but
running ls by hand convinced me that the system could easily be orders
of magnitude slower under some load.
Here are some results, the 2.5.41-mm2v is Con's patch on top of 41-mm2.
The program and some detailed results from the standard test machine are
at http://pages.prodigy.net/davidsen/ if anyone cares. 2.5.41-ac2
crashed every time I tried the benchmark, I'm building the same kernels
on an SMP machine to get a set of SMP results.
2.4.18-5.out
Starting 1 CPU run with 92 MB RAM, minimum 5 data points at 20 sec intervals
_____________ delay ms. ____________
Test low high median average S.D. ratio
noload 238.745 247.882 239.419 240.904 0.004 1.000
smallwrite 259.114 412.260 345.788 326.670 0.056 1.356
largewrite 279.599 2018.226 1155.189 1140.001 0.865 4.732
cpuload 238.958 326.137 239.678 257.048 0.039 1.067
spawnload 227.693 294.703 229.801 246.484 0.029 1.023
8ctx-mem 914.206 26547.601 4402.521 9767.470 10.440 40.545
2ctx-mem 296.088 9228.024 3445.643 4133.895 2.892 17.160
2.5.38-mm2-1.out
Starting 1 CPU run with 92 MB RAM, minimum 5 data points at 20 sec intervals
_____________ delay ms. ____________
Test low high median average S.D. ratio
noload 249.844 251.639 250.326 250.462 0.001 1.000
smallwrite 28972.832 89778.196 69264.628 60491.810 24.170 241.521
largewrite 31667.273 97387.471 34728.573 47237.857 28.243 188.603
cpuload 249.531 674.218 249.920 334.724 0.190 1.336
spawnload 225.724 294.523 225.860 239.641 0.031 0.957
8ctx-mem 1310.153 16910.239 13054.624 11026.391 6.351 44.024
2ctx-mem 3551.369 15704.619 9219.185 9445.208 5.805 37.711
2.5.41-ac1-1.out
Starting 1 CPU run with 91 MB RAM, minimum 5 data points at 20 sec intervals
_____________ delay ms. ____________
Test low high median average S.D. ratio
noload 244.534 250.900 250.526 249.541 0.002 1.000
smallwrite 25594.642 105430.087 64438.179 68740.681 29.479 275.468
largewrite 42475.273 128347.772 65505.753 73467.393 32.260 294.410
cpuload 249.360 1866.355 249.854 573.019 0.723 2.296
spawnload 225.845 296.056 226.379 240.173 0.031 0.962
8ctx-mem 287.816 12466.608 7685.832 6212.891 4.021 24.897
2ctx-mem 2383.892 18277.275 2979.035 6685.271 6.807 26.790
2.5.41-mm2.out
Starting 1 CPU run with 91 MB RAM, minimum 5 data points at 20 sec intervals
_____________ delay ms. ____________
Test low high median average S.D. ratio
noload 251.327 252.137 251.860 251.768 0.000 1.000
smallwrite 281.416 468.787 336.952 353.652 0.069 1.405
largewrite 269.577 2328.442 756.294 937.793 0.748 3.725
cpuload 248.970 330.523 249.991 265.903 0.036 1.056
spawnload 226.116 306.816 227.464 242.903 0.036 0.965
8ctx-mem 4360.954 11087.563 6240.896 6575.003 2.724 26.115
2ctx-mem 4716.368 9847.308 7721.747 7642.746 1.639 30.356
2.5.41-mm2v.out
Starting 1 CPU run with 91 MB RAM, minimum 5 data points at 20 sec intervals
_____________ delay ms. ____________
Test low high median average S.D. ratio
noload 249.779 256.051 249.983 251.590 0.003 1.000
smallwrite 288.109 382.038 348.342 338.549 0.039 1.346
largewrite 276.960 627.196 421.086 413.242 0.146 1.643
cpuload 250.098 329.824 250.630 266.666 0.035 1.060
spawnload 227.521 300.357 228.143 242.750 0.032 0.965
8ctx-mem 1748.992 7175.653 3942.950 4188.664 2.071 16.649
2ctx-mem 2336.009 12493.007 4941.117 6052.204 3.689 24.056
-- bill davidsen <davidsen@tmr.com> CTO, TMR Associates, Inc Doing interesting things with little computers since 1979.- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Oct 15 2002 - 22:00:45 EST