From: Corrado Zoccolo <czoccolo@gmail.com>
To: Mel Gorman <mel@csn.ul.ie>
Cc: Jens Axboe <jens.axboe@oracle.com>,
Andrew Morton <akpm@linux-foundation.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
Frans Pop <elendil@planet.nl>, Jiri Kosina <jkosina@suse.cz>,
Sven Geggus <lists@fuchsschwanzdomain.de>,
Karol Lewandowski <karol.k.lewandowski@gmail.com>,
Tobias Oetiker <tobi@oetiker.ch>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Pekka Enberg <penberg@cs.helsinki.fi>,
Rik van Riel <riel@redhat.com>,
Christoph Lameter <cl@linux-foundation.org>,
Stephan von Krawczynski <skraw@ithnet.com>,
"Rafael J. Wysocki" <rjw@sisk.pl>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH-RFC] cfq: Disable low_latency by default for 2.6.32
Date: Fri, 27 Nov 2009 13:03:29 +0100 [thread overview]
Message-ID: <4e5e476b0911270403n1387dcdco6a36e5923b8e04bc@mail.gmail.com> (raw)
In-Reply-To: <20091127114450.GK13095@csn.ul.ie>
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=UTF-8, Size: 17738 bytes --]
On Fri, Nov 27, 2009 at 12:44 PM, Mel Gorman <mel@csn.ul.ie> wrote:> On Thu, Nov 26, 2009 at 04:18:18PM +0100, Corrado Zoccolo wrote:>> > <SNIP>>> >>> > In case you mean a partial disabling of cfq_latency, I'm try the>> > following patch. The intention is to disable the low_latency logic if>> > kswapd is at work and presumably needs clean pages. Alternative>> > suggestions welcome.>> As it turned out, that patch sucked so I aborted the test and I need to> think about it a lot more.What about using the dirty ratio, instead of checking if kswapd is running?
>> Yes, I meant exactly to disable that part, and doing it when kswapd is>> active is probably a good choice.>> I have a different idea for 2.6.33, though.>> If you have a reliable reproducer of the issue, can you test it on>> git://git.kernel.dk/linux-2.6-block.git branch for-2.6.33?>> It may already be unaffected, since we had various performance>> improvements there, but I think a better way to boost writeback is>> possible.>>>> I haven't tested the high-order allocation scenario yet but the results> as thing stands are below. There are four kernels being compared>> 1. with-low-latency        is 2.6.32-rc8 vanilla> 2. with-low-latency-block-2.6.33  is with the for-2.6.33 from linux-block applied> 3. with-low-latency-async-rampup  is with "[RFC,PATCH] cfq-iosched: improve async queue ramp up formula"> 4. without-low-latency       is with low_latency disabled>> SYSBENCH>         sysbench-with    low-latency    low-latency  sysbench-without>          low-latency    block-2.6.33    async-rampup    low-latency>      1  1266.02 ( 0.00%)  824.08 (-53.63%)  1265.15 (-0.07%)  1278.55 ( 0.98%)>      2  1182.58 ( 0.00%)  1226.42 ( 3.57%)  1223.03 ( 3.31%)  1379.25 (14.26%)>      3  1218.64 ( 0.00%)  1271.38 ( 4.15%)  1246.42 ( 2.23%)  1580.08 (22.87%)>      4  1212.11 ( 0.00%)  1257.84 ( 3.64%)  1325.17 ( 8.53%)  1534.17 (20.99%)>      5  1046.77 ( 0.00%)  981.71 (-6.63%)  1008.44 (-3.80%)  1552.48 (32.57%)>      6  1187.14 ( 0.00%)  1132.89 (-4.79%)  1147.18 (-3.48%)  1661.19 (28.54%)>      7  1179.37 ( 0.00%)  1183.61 ( 0.36%)  1202.49 ( 1.92%)  790.26 (-49.24%)>      8  1164.62 ( 0.00%)  1143.54 (-1.84%)  1184.56 ( 1.68%)  854.10 (-36.36%)>      9  1095.22 ( 0.00%)  1178.72 ( 7.08%)  1002.42 (-9.26%)  1655.04 (33.83%)>      10  1147.52 ( 0.00%)  1153.46 ( 0.52%)  1151.73 ( 0.37%)  1653.89 (30.62%)>      11  823.38 ( 0.00%)  820.64 (-0.33%)  754.15 (-9.18%)  1627.45 (49.41%)>      12  813.73 ( 0.00%)  791.44 (-2.82%)  848.32 ( 4.08%)  1494.63 (45.56%)>      13  898.22 ( 0.00%)  789.63 (-13.75%)  931.47 ( 3.57%)  1521.64 (40.97%)>      14  873.50 ( 0.00%)  938.90 ( 6.97%)  875.75 ( 0.26%)  1311.09 (33.38%)>      15  808.32 ( 0.00%)  979.88 (17.51%)  877.87 ( 7.92%)  1009.70 (19.94%)>      16  758.17 ( 0.00%)  1096.81 (30.87%)  881.23 (13.96%)  725.17 (-4.55%)>> sysbench is helped by both both block-2.6.33 and async-rampup to some> extent. For many of the results, plain old disabling low_latency still> helps the most.>> desktop-net-gitk>           gitk-with    low-latency    low-latency    gitk-without>          low-latency    block-2.6.33    async-rampup    low-latency> min       954.46 ( 0.00%)  570.06 (40.27%)  796.22 (16.58%)  640.65 (32.88%)> mean      964.79 ( 0.00%)  573.96 (40.51%)  798.01 (17.29%)  655.57 (32.05%)> stddev      10.01 ( 0.00%)   2.65 (73.55%)   1.91 (80.95%)   13.33 (-33.18%)> max       981.23 ( 0.00%)  577.21 (41.17%)  800.91 (18.38%)  675.65 (31.14%)>> The changes for block in 2.6.33 make a massive difference here, notably> beating the disabling of low_latency.Yes. These are read of lots of small files, so the improvements forseeky workload we introduced in 2.6.33 helps a lot here.>> IOZone>              iozone-with      low-latency      low-latency     iozone-without>              low-latency      block-2.6.33      async-rampup      low-latency> write-64        151212 ( 0.00%)    163359 ( 7.44%)    163359 ( 7.44%)    159856 ( 5.41%)> write-128        189357 ( 0.00%)    184922 (-2.40%)    202805 ( 6.63%)    206233 ( 8.18%)> write-256        219883 ( 0.00%)    211232 (-4.10%)    189867 (-15.81%)    223174 ( 1.47%)> write-512        224932 ( 0.00%)    222601 (-1.05%)    204459 (-10.01%)    220227 (-2.14%)> write-1024       227738 ( 0.00%)    226728 (-0.45%)    216009 (-5.43%)    226155 (-0.70%)> write-2048       227564 ( 0.00%)    224167 (-1.52%)    229387 ( 0.79%)    224848 (-1.21%)> write-4096       208556 ( 0.00%)    227707 ( 8.41%)    216908 ( 3.85%)    223430 ( 6.66%)> write-8192       219484 ( 0.00%)    222365 ( 1.30%)    217737 (-0.80%)    219389 (-0.04%)> write-16384       206670 ( 0.00%)    209355 ( 1.28%)    204146 (-1.24%)    206295 (-0.18%)> write-32768       203023 ( 0.00%)    205097 ( 1.01%)    199766 (-1.63%)    201852 (-0.58%)> write-65536       162134 ( 0.00%)    196670 (17.56%)    189975 (14.66%)    189173 (14.29%)> write-131072       68534 ( 0.00%)     69145 ( 0.88%)     64519 (-6.22%)     67417 (-1.66%)> write-262144       32936 ( 0.00%)     28587 (-15.21%)     31470 (-4.66%)     27750 (-18.69%)> write-524288       24044 ( 0.00%)     23560 (-2.05%)     23116 (-4.01%)     23759 (-1.20%)> rewrite-64       755681 ( 0.00%)    800767 ( 5.63%)    469931 (-60.81%)    755681 ( 0.00%)> rewrite-128       581518 ( 0.00%)    639723 ( 9.10%)    591774 ( 1.73%)    799840 (27.30%)> rewrite-256       639427 ( 0.00%)    710511 (10.00%)    666414 ( 4.05%)    659861 ( 3.10%)> rewrite-512       669577 ( 0.00%)    743788 ( 9.98%)    692017 ( 3.24%)    684954 ( 2.24%)> rewrite-1024      680960 ( 0.00%)    755195 ( 9.83%)    701422 ( 2.92%)    686182 ( 0.76%)> rewrite-2048      685263 ( 0.00%)    743123 ( 7.79%)    703445 ( 2.58%)    692780 ( 1.09%)> rewrite-4096      631352 ( 0.00%)    686776 ( 8.07%)    640007 ( 1.35%)    643266 ( 1.85%)> rewrite-8192      442146 ( 0.00%)    474089 ( 6.74%)    457768 ( 3.41%)    442624 ( 0.11%)> rewrite-16384      428641 ( 0.00%)    454857 ( 5.76%)    442896 ( 3.22%)    432613 ( 0.92%)> rewrite-32768      425361 ( 0.00%)    444206 ( 4.24%)    434472 ( 2.10%)    430568 ( 1.21%)> rewrite-65536      405183 ( 0.00%)    433898 ( 6.62%)    419843 ( 3.49%)    389242 (-4.10%)> rewrite-131072      66110 ( 0.00%)     58370 (-13.26%)     54342 (-21.66%)     58472 (-13.06%)> rewrite-262144      29254 ( 0.00%)     24665 (-18.61%)     25710 (-13.78%)     29306 ( 0.18%)> rewrite-524288      23812 ( 0.00%)     20742 (-14.80%)     22490 (-5.88%)     24543 ( 2.98%)> read-64         934589 ( 0.00%)    1160938 (19.50%)    1004538 ( 6.96%)    840903 (-11.14%)> read-128        1601534 ( 0.00%)    1869179 (14.32%)    1681806 ( 4.77%)    1280633 (-25.06%)> read-256        1255511 ( 0.00%)    1526887 (17.77%)    1304314 ( 3.74%)    1310683 ( 4.21%)> read-512        1291158 ( 0.00%)    1377278 ( 6.25%)    1336145 ( 3.37%)    1319723 ( 2.16%)> read-1024       1319408 ( 0.00%)    1306564 (-0.98%)    1368162 ( 3.56%)    1347557 ( 2.09%)> read-2048       1316016 ( 0.00%)    1394645 ( 5.64%)    1339827 ( 1.78%)    1347393 ( 2.33%)> read-4096       1253710 ( 0.00%)    1307525 ( 4.12%)    1247519 (-0.50%)    1251882 (-0.15%)> read-8192        995149 ( 0.00%)    1033337 ( 3.70%)    1016944 ( 2.14%)    1011794 ( 1.65%)> read-16384       883156 ( 0.00%)    905213 ( 2.44%)    905213 ( 2.44%)    897458 ( 1.59%)> read-32768       844368 ( 0.00%)    855213 ( 1.27%)    849609 ( 0.62%)    856364 ( 1.40%)> read-65536       816099 ( 0.00%)    839262 ( 2.76%)    835019 ( 2.27%)    826473 ( 1.26%)> read-131072       818055 ( 0.00%)    837369 ( 2.31%)    828230 ( 1.23%)    824351 ( 0.76%)> read-262144       827225 ( 0.00%)    839635 ( 1.48%)    840538 ( 1.58%)    835693 ( 1.01%)> read-524288       24653 ( 0.00%)     21387 (-15.27%)     20602 (-19.66%)     22519 (-9.48%)> reread-64       2329708 ( 0.00%)    2251544 (-3.47%)    1985134 (-17.36%)    1985134 (-17.36%)> reread-128       1446222 ( 0.00%)    1979446 (26.94%)    2009076 (28.02%)    2137031 (32.33%)> reread-256       1828508 ( 0.00%)    2006158 ( 8.86%)    1892980 ( 3.41%)    1879725 ( 2.72%)> reread-512       1521718 ( 0.00%)    1642783 ( 7.37%)    1508887 (-0.85%)    1579934 ( 3.68%)> reread-1024      1347557 ( 0.00%)    1422540 ( 5.27%)    1384034 ( 2.64%)    1375171 ( 2.01%)> reread-2048      1340664 ( 0.00%)    1413929 ( 5.18%)    1372364 ( 2.31%)    1350783 ( 0.75%)> reread-4096      1259592 ( 0.00%)    1324868 ( 4.93%)    1273788 ( 1.11%)    1284839 ( 1.96%)> reread-8192      1007285 ( 0.00%)    1033710 ( 2.56%)    1027159 ( 1.93%)    1011317 ( 0.40%)> reread-16384      891404 ( 0.00%)    910828 ( 2.13%)    916562 ( 2.74%)    905022 ( 1.50%)> reread-32768      850492 ( 0.00%)    859341 ( 1.03%)    856385 ( 0.69%)    862772 ( 1.42%)> reread-65536      836565 ( 0.00%)    852664 ( 1.89%)    852315 ( 1.85%)    847020 ( 1.23%)> reread-131072      844516 ( 0.00%)    862590 ( 2.10%)    854067 ( 1.12%)    853155 ( 1.01%)> reread-262144      851524 ( 0.00%)    860559 ( 1.05%)    864921 ( 1.55%)    860653 ( 1.06%)> reread-524288      24927 ( 0.00%)     21300 (-17.03%)     19748 (-26.23%)     22487 (-10.85%)> randread-64      1605256 ( 0.00%)    1605256 ( 0.00%)    1605256 ( 0.00%)    1775099 ( 9.57%)> randread-128      1179358 ( 0.00%)    1582649 (25.48%)    1511363 (21.97%)    1528576 (22.85%)> randread-256      1421755 ( 0.00%)    1599680 (11.12%)    1460430 ( 2.65%)    1310683 (-8.47%)> randread-512      1306873 ( 0.00%)    1278855 (-2.19%)    1243315 (-5.11%)    1281909 (-1.95%)> randread-1024     1201314 ( 0.00%)    1254656 ( 4.25%)    1190657 (-0.90%)    1231629 ( 2.46%)> randread-2048     1179413 ( 0.00%)    1227971 ( 3.95%)    1185272 ( 0.49%)    1190529 ( 0.93%)> randread-4096     1107005 ( 0.00%)    1160862 ( 4.64%)    1110727 ( 0.34%)    1116792 ( 0.88%)> randread-8192      894337 ( 0.00%)    924264 ( 3.24%)    912676 ( 2.01%)    899487 ( 0.57%)> randread-16384     783760 ( 0.00%)    800299 ( 2.07%)    793351 ( 1.21%)    791341 ( 0.96%)> randread-32768     740498 ( 0.00%)    743720 ( 0.43%)    741233 ( 0.10%)    743511 ( 0.41%)> randread-65536     721640 ( 0.00%)    727692 ( 0.83%)    726984 ( 0.74%)    728139 ( 0.89%)> randread-131072     715284 ( 0.00%)    722094 ( 0.94%)    717746 ( 0.34%)    720825 ( 0.77%)> randread-262144     709855 ( 0.00%)    706770 (-0.44%)    709133 (-0.10%)    714943 ( 0.71%)> randread-524288      394 ( 0.00%)      421 ( 6.41%)      418 ( 5.74%)      431 ( 8.58%)> randwrite-64      730988 ( 0.00%)    764288 ( 4.36%)    723111 (-1.09%)    730988 ( 0.00%)> randwrite-128      746459 ( 0.00%)    799840 ( 6.67%)    746459 ( 0.00%)    742331 (-0.56%)> randwrite-256      695778 ( 0.00%)    752329 ( 7.52%)    720041 ( 3.37%)    727850 ( 4.41%)> randwrite-512      666253 ( 0.00%)    722760 ( 7.82%)    667081 ( 0.12%)    691126 ( 3.60%)> randwrite-1024     651223 ( 0.00%)    697776 ( 6.67%)    663292 ( 1.82%)    659625 ( 1.27%)> randwrite-2048     655558 ( 0.00%)    691887 ( 5.25%)    665720 ( 1.53%)    664073 ( 1.28%)> randwrite-4096     635556 ( 0.00%)    662721 ( 4.10%)    643170 ( 1.18%)    642400 ( 1.07%)> randwrite-8192     467357 ( 0.00%)    491364 ( 4.89%)    476720 ( 1.96%)    469734 ( 0.51%)> randwrite-16384     413188 ( 0.00%)    427521 ( 3.35%)    417353 ( 1.00%)    417282 ( 0.98%)> randwrite-32768     404161 ( 0.00%)    411721 ( 1.84%)    404942 ( 0.19%)    407580 ( 0.84%)> randwrite-65536     379372 ( 0.00%)    397312 ( 4.52%)    386853 ( 1.93%)    381273 ( 0.50%)> randwrite-131072     21780 ( 0.00%)     16924 (-28.69%)     21177 (-2.85%)     19758 (-10.23%)> randwrite-262144     6249 ( 0.00%)     5548 (-12.64%)     6370 ( 1.90%)     6316 ( 1.06%)> randwrite-524288     2915 ( 0.00%)     2582 (-12.90%)     2871 (-1.53%)     2859 (-1.96%)> bkwdread-64      1141196 ( 0.00%)    1141196 ( 0.00%)    1004538 (-13.60%)    1141196 ( 0.00%)> bkwdread-128      1066865 ( 0.00%)    1386465 (23.05%)    1400936 (23.85%)    1101900 ( 3.18%)> bkwdread-256      877797 ( 0.00%)    1105556 (20.60%)    1105556 (20.60%)    1105556 (20.60%)> bkwdread-512      1133103 ( 0.00%)    1162547 ( 2.53%)    1175271 ( 3.59%)    1162547 ( 2.53%)> bkwdread-1024     1163562 ( 0.00%)    1206714 ( 3.58%)    1213534 ( 4.12%)    1195962 ( 2.71%)> bkwdread-2048     1163439 ( 0.00%)    1218910 ( 4.55%)    1204552 ( 3.41%)    1204552 ( 3.41%)> bkwdread-4096     1116792 ( 0.00%)    1175477 ( 4.99%)    1159922 ( 3.72%)    1150600 ( 2.94%)> bkwdread-8192      912288 ( 0.00%)    935233 ( 2.45%)    944695 ( 3.43%)    934724 ( 2.40%)> bkwdread-16384     817707 ( 0.00%)    824140 ( 0.78%)    832527 ( 1.78%)    829152 ( 1.38%)> bkwdread-32768     775898 ( 0.00%)    773714 (-0.28%)    785494 ( 1.22%)    787691 ( 1.50%)> bkwdread-65536     759643 ( 0.00%)    769924 ( 1.34%)    778780 ( 2.46%)    772174 ( 1.62%)> bkwdread-131072     763215 ( 0.00%)    769634 ( 0.83%)    773707 ( 1.36%)    773816 ( 1.37%)> bkwdread-262144     765491 ( 0.00%)    768992 ( 0.46%)    780876 ( 1.97%)    780021 ( 1.86%)> bkwdread-524288      3688 ( 0.00%)     3595 (-2.59%)     3577 (-3.10%)     3724 ( 0.97%)>> The upcoming changes for 2.6.33 also help iozone in many cases, often by more> than just disabling low_latency. It has the occasional massive gain or loss> for the larger file sizes. I don't know why this is but as the big losses> appear to be mostly in the write-tests, I would guess that it's differences> in heavy-writer-throttling.I wonder if 2.6.33 + my async rampup patch will improve still further,maybe reaching the low_latency=0 performance also for writing tests.>> The only downside with block-2.6.33 is that there are a lot of patches in> there and doesn't help with the 2.6.32 release as such. I could do a reverse> bisect to see what helps the most in there but under ideal conditions, it'll> take 3 days to complete and I wouldn't be able to start until Monday as I'm> out of the country for the weekend. That's a bit late.Bisect will likely not help, since we have several patch series withheavy internal dependencies in that tree.If one of the patch series is found to bring the improvement, you haveto backport the entire series, that is not advisable for a rc8 or forstable.>> p.s. As a consequence of being out of the country, I also won't be able to>   respond to mail over the weekend.>> --> Mel Gorman>Thanks for the detailed reportCorradoÿôèº{.nÇ+·®+%Ëÿ±éݶ\x17¥wÿº{.nÇ+·¥{±þG«éÿ{ayº\x1dÊÚë,j\a¢f£¢·hïêÿêçz_è®\x03(éÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?¨èÚ&£ø§~á¶iOæ¬z·vØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?I¥
next prev parent reply other threads:[~2009-11-27 12:03 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-11-26 12:19 Mel Gorman
2009-11-26 13:08 ` Mike Galbraith
2009-11-26 13:20 ` Bartlomiej Zolnierkiewicz
2009-11-26 13:37 ` Mike Galbraith
2009-11-26 13:56 ` Mel Gorman
2009-11-26 13:47 ` Corrado Zoccolo
2009-11-26 14:17 ` Mel Gorman
2009-11-26 15:18 ` Corrado Zoccolo
2009-11-27 11:44 ` Mel Gorman
2009-11-27 12:03 ` Corrado Zoccolo [this message]
2009-11-27 15:58 ` Mel Gorman
2009-11-27 18:14 ` Corrado Zoccolo
2009-11-27 18:52 ` Mel Gorman
2009-11-29 15:11 ` Corrado Zoccolo
2009-11-30 12:04 ` Mel Gorman
2009-11-30 12:54 ` Corrado Zoccolo
2009-11-30 15:48 ` Mel Gorman
2009-11-30 17:21 ` Corrado Zoccolo
2009-11-27 5:58 ` KOSAKI Motohiro
2009-11-27 6:29 ` KOSAKI Motohiro
2009-11-27 12:16 ` Mel Gorman
2009-11-30 10:18 ` KOSAKI Motohiro
2009-11-27 4:36 ` KOSAKI Motohiro
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4e5e476b0911270403n1387dcdco6a36e5923b8e04bc@mail.gmail.com \
--to=czoccolo@gmail.com \
--cc=akpm@linux-foundation.org \
--cc=cl@linux-foundation.org \
--cc=elendil@planet.nl \
--cc=jens.axboe@oracle.com \
--cc=jkosina@suse.cz \
--cc=karol.k.lewandowski@gmail.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=lists@fuchsschwanzdomain.de \
--cc=mel@csn.ul.ie \
--cc=penberg@cs.helsinki.fi \
--cc=riel@redhat.com \
--cc=rjw@sisk.pl \
--cc=skraw@ithnet.com \
--cc=tobi@oetiker.ch \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Powered by JetHome