On Tuesday 23 October 2001 01:11, Robert Love wrote:
> On Mon, 2001-10-22 at 16:43, elko wrote:
> > My current conclusion: this combination of kernel and patches
> > is the most responsive I've ever used, normally, when I run
> > these command's, my systems would freeze to the point I had
> > to give them the VNP.
>
> Excellent. Give a lot of credit to Rik too because his VM work has
> relieved a lot of thrashing/storming in high-load situations such as
> yours.
> Robert Love
Hereby, I also give *a lot of* credit to Rik, by CC'ing him ;^)
(Hi Rik! No worries, I'm not from Vries <g>)
This system has not been running as smooth as now, since I switched
from 2.4.0 to 2.4.10-ac12 and a few days later to 2.4.12-ac3,
so I did some testing, that I could never complete with
the previous kernels I used (as was asked on the list).
I've attached the results of my latest tests..
-- ElkOS: 4:34pm up 1:05, 3 users, load average: 2.61, 2.58, 2.83 bofhX: Network failure - call NBC--------------Boundary-00=_JMZNCXHEFIC99GNHLE3E Content-Type: text/plain; charset="iso-8859-1"; name="tests" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename="tests"
Tests done just after a reboot:
test1: In console, play a mp3-file with mpg123 test2: In console, play a mp3-file with mpg123 and run the find command test3: Run test1 in X, play with freeamp test4: Run test2 in X, play with freeamp test5: Run test2 in X, play with mpg123 no reboot after test 4, since memory was released, a bit of swap stayed behind... read the output below...
Extra apps in X: gkrellm, 2x setiathome, 4 konsoles, freeamp playing, kmail, 2x konqueror, licq
- Fire up mpg123: [elko@ElkOS mp3]$ while :; do mpg123 "Aphex Twin - Come to Daddy.mp3"; do= ne
- `find' command to stress the system (~3,7G, ~81946 files): [elko@ElkOS ~]# find .|xargs slocate|sort|uniq -c|head -1
- Make sure I'm using /dev/dsp: [elko@ElkOS ~]# /usr/sbin/lsof|grep dsp esd 605 elko 5w CHR 14,3 22742 /dev/dsp
- Watch the system: [elko@ElkOS ~]# vmstat -n 60 15 |tee -a ~elko/tests =2E..
- Audio driver used (in kernel): ES1371
- System info: [elko@ElkOS ~]# /sbin/swapon -s Filename Type Size Used Priority /dev/hda7 partition 104380 33648 -1 /dev/hdd5 partition 1465592 35144 -2
[elko@ElkOS ~]# dmesg | egrep "clock |Mem" Memory: 577440k/589824k available \ (1177k kernel code, 12000k reserved, \ 347k data, 236k init, 0k highmem) =2E.... CPU clock speed is 852.0020 MHz. =2E.... host bus clock speed is 100.2353 MHz.
* Now for the tests: -------------------------------------
test1: 2.4.12-ac3-vmpatch-freeswap-preempt
- No skips in mp3 playing, nothing special to report
- vmstat output: procs memory swap io system c= pu r b w swpd free buff cache si so bi bo in cs us sy = id 1 1 0 0 455808 83428 13192 0 0 341 4 220 262 1 3 = 96 1 0 0 0 454752 83428 14152 0 0 16 2 792 1390 2 1 = 97 1 0 0 0 453768 83428 15112 0 0 16 1 790 1388 2 1 = 97 1 0 0 0 452700 83428 16136 0 0 17 0 790 1387 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 14 0 790 1386 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 1 789 1385 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 0 789 1386 2 1 = 97 1 0 0 0 451852 83428 16948 0 0 0 0 789 1386 2 1 = 97 1 0 0 0 451852 83428 16948 0 0 0 0 789 1386 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 1 789 1386 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 0 789 1386 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 0 789 1386 2 1 = 97 1 0 0 0 451852 83428 16948 0 0 0 0 789 1386 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 0 789 1387 2 1 = 97 1 0 0 0 451856 83428 16948 0 0 0 1 789 1386 2 1 = 97 -------------------------------------
test2: 2.4.12-ac3-vmpatch-freeswap-preempt
- This stressed my system a little, she couldn't handle the load perfect, but she managed to stay alive ;^)
- Top1 when vmstat was finished, but the find command still running (top)= : 658 elko 19 0 766M 321M 536 R 321M 98.0 57.0 1:01 slocat= e
- Uptimes shows no big load (uptime): 2:18pm up 9 min, 4 users, load average: 1.45, 0.85, 0.38 2:19pm up 9 min, 4 users, load average: 1.47, 0.91, 0.41 2:21pm up 12 min, 4 users, load average: 1.56, 1.14, 0.57 2:27pm up 18 min, 4 users, load average: 1.45, 1.34, 0.84
- Memory suddenly released (free): total used free shared buffers cached Mem: 577676 451124 126552 0 168 7124 -/+ buffers/cache: 443832 133844 Swap: 1569972 6696 1563276
- Skips of mp3-file during test: 1st 5 mins.: 5 skips ~1,5 second each 2nd 5 mins.: 14 skips, same length 3rd 5 mins.: 15 skips, same length
- vmstat output: procs memory swap io system = cpu r b w swpd free buff cache si so bi bo in cs us sy = id 1 0 0 0 445268 83860 17560 0 0 308 4 186 199 2 3 = 95 2 0 0 356740 3064 228 6372 3 3950 18 4054 820 1473 74 19 = 8 2 0 0 6904 266992 244 6196 15 4692 39 4692 806 1295 77 17 = 6 2 0 0 6908 366392 280 6116 15 4190 39 4191 810 1291 76 19 = 6 2 0 0 557888 3064 140 5816 25 4617 61 4620 783 1433 70 18 = 12 2 0 0 14200 454988 224 6704 31 9375 60 9377 771 1116 66 24 = 10 2 0 0 451308 2860 120 5856 9 3477 26 3477 769 1409 75 16 = 9 2 0 0 946388 3900 124 5900 45 10954 75 10963 797 1170 62 21 = 18 2 0 0 492616 2860 100 5520 16 4142 39 4142 795 1267 75 18 = 7 2 0 0 979576 3060 104 5172 8 9130 26 9130 815 1174 70 25 = 5 2 0 0 450700 3064 108 5352 27 1696 72 1697 799 1313 73 15 = 12 2 0 0 860220 2844 104 5212 6 10968 22 10968 790 1386 64 22 = 14 2 0 0 445292 2588 96 5448 13 3835 43 3835 869 1399 71 16 = 12 2 0 0 895300 3060 104 4988 6 7457 23 7457 787 1187 70 21 = 9 2 0 0 443772 3064 164 4752 37 4832 86 4834 808 1277 68 17 = 15 -------------------------------------
test3: 2.4.12-ac3-vmpatch-freeswap-preempt
- No skips in mp3 playing, nothing special to report
- Using /dev/dsp with freeamp: [elko@ElkOS ~]$ /usr/sbin/lsof | grep dsp freeamp 716 elko 17w CHR 14,3 22742 /dev/dsp freeamp 717 elko 17w CHR 14,3 22742 /dev/dsp freeamp 718 elko 17w CHR 14,3 22742 /dev/dsp freeamp 719 elko 17w CHR 14,3 22742 /dev/dsp freeamp 720 elko 17w CHR 14,3 22742 /dev/dsp freeamp 721 elko 17w CHR 14,3 22742 /dev/dsp freeamp 723 elko 17w CHR 14,3 22742 /dev/dsp freeamp 852 elko 17w CHR 14,3 22742 /dev/dsp freeamp 853 elko 17w CHR 14,3 22742 /dev/dsp freeamp 854 elko 17w CHR 14,3 22742 /dev/dsp
- vmstat output: procs memory swap io system cpu r b w swpd free buff cache si so bi bo in cs us sy id 4 0 0 0 301868 86072 62260 0 0 114 5 138 280 73 8 19 2 0 0 0 305256 86072 62680 0 0 8 3 303 752 90 10 0 2 0 0 0 302248 86084 63716 0 0 17 4 316 575 88 12 0 7 0 0 0 298572 86104 64876 0 0 17 9 296 507 90 10 0 4 0 0 0 297036 86108 65976 0 0 18 5 311 453 91 9 0 2 0 0 0 295916 86112 66916 0 0 16 3 300 523 88 12 0 4 0 0 0 295220 86112 67812 0 0 15 1 274 322 92 8 0 2 0 0 0 293344 86112 68728 0 0 15 1 276 344 93 7 0 3 0 0 0 291392 86184 70644 0 0 33 5 286 403 90 10 0 3 0 0 0 290636 86196 71588 0 0 16 1 296 551 89 11 0 2 0 0 0 291744 86196 72484 0 0 15 1 278 371 92 8 0 2 0 0 0 290828 86196 73380 0 0 15 1 282 389 93 7 0 2 0 0 0 290024 86200 74168 0 0 13 0 279 385 89 11 0 2 0 0 0 287168 86212 75412 0 0 21 2 291 509 91 9 0 2 0 0 0 286228 86212 76308 0 0 15 2 274 332 92 8 0 -------------------------------------
test4: 2.4.12-ac3-vmpatch-freeswap-preempt
- This is a strange one, since I've loaded more apps. in X, I would think= the system would be even more stressed then in test 2 (same test in console= ); maybe it's freeamp that's doing so well, since I only had 2 skips; therefore, I'll run this test again with mpg123, like in test2
- Skips of mp3-file during test (4 skips total): 1st 5 mins.: none 2nd 5 mins.: none 3rd 5 mins.: 1 at 12 mins. and 1 at ~14,9678 mins.
- Note: 3 times, when memory began to get full, swap took over without sl= owing anything down at all, you couldn't notice it...
- vmstat output: procs memory swap io system = cpu r b w swpd free buff cache si so bi bo in cs us sy = id 3 2 0 0 281596 100628 61968 0 0 218 6 176 309 26 5 = 69 4 0 0 0 225064 108032 81312 0 0 403 26 368 516 85 15 = 0 3 0 0 0 116020 113436 96648 0 0 303 62 347 484 88 12 = 0 4 0 0 212492 3068 103368 32608 0 0 42 20 280 333 86 14 = 0 3 3 0 327916 3060 776 21440 13 4180 47 4183 322 331 83 17 = 0 3 0 0 487860 3064 636 14908 89 3045 121 3054 302 309 86 14 = 0 3 0 0 750580 3060 384 12788 36 3023 52 3024 301 315 82 18 = 0 3 0 0 68756 292060 704 18208 131 0 226 3 277 347 86 14 = 0 3 0 0 68756 31500 704 19232 3 0 20 1 273 309 88 12 = 0 3 0 0 472564 3064 424 13860 34 3912 68 3913 314 377 77 17 = 7 3 0 0 68992 318612 812 20552 415 1911 555 1913 291 357 86 14 = 0 3 0 0 68960 63564 988 22088 15 0 43 1 277 318 88 12 = 0 3 0 0 480228 2552 476 13688 37 3446 67 3447 292 325 76 20 = 4 4 0 0 551760 3120 584 14076 281 3791 333 3792 324 357 84 15 = 1 3 0 0 863540 2896 428 8616 14 4522 32 4523 305 309 76 17 = 7 -------------------------------------
test5: 2.4.12-ac3-vmpatch-freeswap-preempt
- Some skips, but not as much as I would have thought: 1st skip at 7 mins. (during a desktop-switch) 2nd skip at 9,5 mins. 3rd skip at 13 mins. 4th skip at 14 mins. (just before the end of vmstat)
- I would have expected much more skips during this test, since it's the same as test2, only much more programs running (X etc.)...
- When running these tests in X, the machine would freeze when trying to switch kde-desktops or switch apps., but when nothing is touched, she just (about) keeps on playing...
- vmstat output: procs memory swap io system = cpu r b w swpd free buff cache si so bi bo in cs us sy = id 5 1 0 68120 465408 1188 24364 45 766 161 773 216 317 66 10 = 24 4 0 0 68120 412296 8892 41464 5 0 384 27 884 1508 88 12 = 0 4 0 0 68120 316464 14652 58128 1 0 323 53 868 1477 88 12 = 0 4 0 0 68120 63152 15636 59784 0 0 44 28 798 1364 85 15 = 0 5 1 0 440360 3064 1132 13772 15 3964 38 3964 847 1335 81 19 = 0 4 0 0 487248 3064 916 21280 381 1927 552 1938 817 1358 85 15 = 0 4 0 0 815376 3064 684 19176 274 3886 386 3889 837 1448 81 19 = 0 4 0 0 70248 259996 924 22368 134 0 190 4 801 1372 86 14 = 0 6 0 0 70208 3064 988 24160 18 0 57 0 796 1421 87 13 = 0 4 0 0 467696 2804 508 15660 30 4383 68 4384 784 1246 74 21 = 5 4 0 0 70356 322744 648 14424 8 1546 29 1549 807 1345 85 15 = 0 4 0 0 70352 60732 872 15580 11 0 34 1 792 1372 87 13 = 0 4 0 0 505900 3064 392 11020 33 4972 95 4973 793 1237 79 18 = 3 4 0 0 568404 3064 520 10104 13 2317 35 2317 785 1298 85 15 = 0 4 0 0 858932 3064 564 10796 43 3068 93 3069 767 1235 79 21 = 0 -------------------------------------
BTW: the `find' command never got to finish... should I let it and `time -v' it for fun ????
~~ End of testing 2.4.12-ac3-vmpatch-freeswap-preempt (23-10-2001 16:51) ~~
-- elko--------------Boundary-00=_JMZNCXHEFIC99GNHLE3E-- - 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/