| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 29 Oct 2019 10:20:57 +0100
From: Dario Faggioli <dfaggioli@...e.com>
To: Aaron Lu <aaron.lu@...ux.alibaba.com>,
Aubrey Li <aubrey.intel@...il.com>
Cc: Tim Chen <tim.c.chen@...ux.intel.com>,
Julien Desfossez <jdesfossez@...italocean.com>,
"Li, Aubrey" <aubrey.li@...ux.intel.com>,
Subhra Mazumdar <subhra.mazumdar@...cle.com>,
Vineeth Remanan Pillai <vpillai@...italocean.com>,
Nishanth Aravamudan <naravamudan@...italocean.com>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...nel.org>,
Thomas Gleixner <tglx@...utronix.de>,
Paul Turner <pjt@...gle.com>,
Linus Torvalds <torvalds@...ux-foundation.org>,
Linux List Kernel Mailing <linux-kernel@...r.kernel.org>,
Frédéric Weisbecker <fweisbec@...il.com>,
Kees Cook <keescook@...omium.org>,
Greg Kerr <kerrnel@...gle.com>, Phil Auld <pauld@...hat.com>,
Valentin Schneider <valentin.schneider@....com>,
Mel Gorman <mgorman@...hsingularity.net>,
Pawan Gupta <pawan.kumar.gupta@...ux.intel.com>,
Paolo Bonzini <pbonzini@...hat.com>
Subject: Re: [RFC PATCH v3 00/16] Core scheduling v3
On Tue, 2019-10-29 at 10:11 +0100, Dario Faggioli wrote:
> On Sun, 2019-09-15 at 22:14 +0800, Aaron Lu wrote:
> > I'm using the following branch as base which is v5.1.5 based:
> > https://github.com/digitalocean/linux-coresched coresched-v3-
> > v5.1.5-
> > test
> >
> > And I have pushed Tim's branch to:
> > https://github.com/aaronlu/linux coresched-v3-v5.1.5-test-tim
> >
> > Mine:
> > https://github.com/aaronlu/linux coresched-v3-v5.1.5-test-
> > core_vruntime
> >
> Hello,
>
> As anticipated, I've been trying to follow the development of this
> feature and, in the meantime, I have done some benchmarks.
>
> I actually have a lot of data (and am planning for more), so I am
> sending a few emails, each one with a subset of the numbers in it,
> instead than just one which would be beyond giant! :-)
>
NETPERF-UNIX
============
http://xenbits.xen.org/people/dariof/benchmarks/results/linux/core-sched/mmtests/boxes/wayrath/coresched-email-7_mutilate.txt
v v BM BM BM BM BM BM
BM-HT BM-noHT HT noHT csc-HT csc_stallfix-HT csc_tim-HT csc_vruntime-HT
Hmean 64 984.93 ( 0.00%) 1011.24 ( 2.67%) 1000.82 ( 1.61%) 1001.98 ( 1.73%) 947.61 ( -3.79%) 823.10 * -16.43%* 789.89 * -19.80%* 928.11 * -5.77%*
Hmean 256 3683.78 ( 0.00%) 3763.52 ( 2.16%) 3788.96 ( 2.86%) 3725.17 ( 1.12%) 1254.25 * -65.95%* 1261.92 * -65.74%* 1264.02 * -65.69%* 1260.30 * -65.79%*
Hmean 2048 7928.28 ( 0.00%) 7845.97 ( -1.04%) 7911.88 ( -0.21%) 7809.11 * -1.50%* 5334.65 * -32.71%* 5340.97 * -32.63%* 5337.93 * -32.67%* 5394.57 * -31.96%*
Hmean 8192 8134.23 ( 0.00%) 8096.88 ( -0.46%) 8258.84 ( 1.53%) 8076.36 ( -0.71%) 5374.33 * -33.93%* 5394.12 * -33.69%* 5504.00 * -32.34%* 5447.92 * -33.02%*
Stddev 64 30.46 ( 0.00%) 31.07 ( -2.00%) 15.36 ( 49.58%) 41.51 ( -36.26%) 54.71 ( -79.62%) 121.84 (-299.98%) 81.75 (-168.35%) 33.75 ( -10.79%)
Stddev 256 102.26 ( 0.00%) 104.86 ( -2.55%) 107.90 ( -5.52%) 116.17 ( -13.61%) 3.32 ( 96.75%) 5.79 ( 94.34%) 14.69 ( 85.63%) 19.37 ( 81.06%)
Stddev 2048 94.73 ( 0.00%) 48.12 ( 49.21%) 137.50 ( -45.15%) 43.30 ( 54.29%) 58.65 ( 38.08%) 50.97 ( 46.20%) 57.43 ( 39.38%) 36.39 ( 61.58%)
Stddev 8192 172.77 ( 0.00%) 48.68 ( 71.83%) 261.65 ( -51.45%) 76.30 ( 55.83%) 40.65 ( 76.47%) 27.61 ( 84.02%) 65.26 ( 62.23%) 31.66 ( 81.68%)
v v VM VM VM VM VM VM
VM-HT VM-noHT HT noHT csc-HT csc_stallfix-HT csc_tim-HT csc_vruntime-HT
Hmean 64 516.67 ( 0.00%) 585.50 ( 13.32%) 592.54 ( 14.68%) 582.70 ( 12.78%) 591.32 ( 14.45%) 546.68 ( 5.81%) 602.58 ( 16.63%) 729.10 * 41.12%*
Hmean 256 1070.01 ( 0.00%) 1306.95 * 22.14%* 1193.89 * 11.58%* 1271.17 * 18.80%* 1362.78 * 27.36%* 1171.49 ( 9.48%) 1335.98 * 24.86%* 1248.79 * 16.71%*
Hmean 2048 5002.14 ( 0.00%) 6865.72 * 37.26%* 5569.42 ( 11.34%) 5074.48 ( 1.45%) 5849.11 ( 16.93%) 4745.97 ( -5.12%) 6330.87 ( 26.56%) 6418.51 ( 28.32%)
Hmean 8192 5116.24 ( 0.00%) 7494.15 * 46.48%* 6960.17 * 36.04%* 6009.31 ( 17.46%) 6114.30 ( 19.51%) 5226.13 ( 2.15%) 6316.40 ( 23.46%) 7924.66 * 54.89%*
Stddev 64 81.30 ( 0.00%) 139.96 ( -72.15%) 162.38 ( -99.72%) 113.29 ( -39.35%) 150.34 ( -84.91%) 162.05 ( -99.32%) 163.74 (-101.39%) 47.94 ( 41.04%)
Stddev 256 64.89 ( 0.00%) 130.57 (-101.20%) 115.02 ( -77.23%) 120.49 ( -85.68%) 106.63 ( -64.31%) 140.18 (-116.01%) 118.29 ( -82.28%) 133.66 (-105.96%)
Stddev 2048 779.45 ( 0.00%) 767.31 ( 1.56%) 1869.81 (-139.89%) 1265.84 ( -62.40%) 1249.59 ( -60.32%) 506.63 ( 35.00%) 1427.22 ( -83.11%) 2296.29 (-194.60%)
Stddev 8192 942.88 ( 0.00%) 2559.80 (-171.49%) 1207.26 ( -28.04%) 1776.59 ( -88.42%) 1405.17 ( -49.03%) 1379.48 ( -46.30%) 2565.86 (-172.13%) 1066.06 ( -13.06%)
v v VM VM VM VM VM VM
VM-v4-HT VM-v4-noHT v4-HT v4-noHT v4-csc-HT v4-csc_stallfix-HT v4-csc_tim-HT v4-csc_vruntime-HT
Hmean 64 626.51 ( 0.00%) 535.18 ( -14.58%) 610.07 ( -2.62%) 509.04 ( -18.75%) 552.16 ( -11.87%) 471.44 * -24.75%* 484.50 * -22.67%* 488.32 * -22.06%*
Hmean 256 999.57 ( 0.00%) 1159.65 * 16.02%* 1209.25 * 20.98%* 1217.94 * 21.85%* 1196.13 * 19.66%* 1286.01 * 28.66%* 1154.57 * 15.51%* 1238.40 * 23.89%*
Hmean 2048 3882.52 ( 0.00%) 4483.92 ( 15.49%) 4969.62 * 28.00%* 4910.98 * 26.49%* 4646.33 * 19.67%* 5247.76 * 35.16%* 4515.47 * 16.30%* 5096.38 * 31.26%*
Hmean 8192 4086.48 ( 0.00%) 4935.20 ( 20.77%) 4711.62 * 15.30%* 5067.04 ( 24.00%) 5887.99 * 44.08%* 5360.18 * 31.17%* 5847.04 * 43.08%* 5990.50 * 46.59%*
Stddev 64 134.26 ( 0.00%) 117.95 ( 12.15%) 67.29 ( 49.88%) 88.91 ( 33.78%) 78.73 ( 41.36%) 22.34 ( 83.36%) 45.36 ( 66.22%) 64.62 ( 51.87%)
Stddev 256 36.69 ( 0.00%) 32.94 ( 10.22%) 93.79 (-155.60%) 52.76 ( -43.79%) 72.03 ( -96.31%) 94.69 (-158.06%) 32.05 ( 12.65%) 62.31 ( -69.82%)
Stddev 2048 64.82 ( 0.00%) 785.16 (-1111.23%) 1086.67 (-1576.36%) 863.76 (-1232.48%) 552.05 (-751.62%) 597.66 (-821.99%) 300.86 (-364.12%) 1057.90 (-1531.98%)
Stddev 8192 248.29 ( 0.00%) 1345.78 (-442.02%) 636.92 (-156.53%) 1497.43 (-503.10%) 1528.17 (-515.48%) 204.43 ( 17.66%) 788.65 (-217.63%) 1380.18 (-455.88%)
v v VMx2 VMx2 VMx2 VMx2 VMx2 VMx2
VMx2-HT VMx2-noHT HT noHT csc-HT csc_stallfix-HT csc_tim-HT csc_vruntime-HT
Hmean 64 575.39 ( 0.00%) 230.57 * -59.93%* 525.97 ( -8.59%) 241.09 * -58.10%* 671.83 ( 16.76%) 574.64 ( -0.13%) 676.93 ( 17.65%) 713.81 ( 24.06%)
Hmean 256 1243.12 ( 0.00%) 679.95 * -45.30%* 1262.76 ( 1.58%) 646.82 * -47.97%* 1607.80 * 29.34%* 1297.86 ( 4.40%) 1573.09 * 26.54%* 1244.30 ( 0.09%)
Hmean 2048 4448.89 ( 0.00%) 3020.71 * -32.10%* 4460.65 ( 0.26%) 3342.89 * -24.86%* 7086.92 * 59.30%* 4544.81 ( 2.16%) 4209.05 ( -5.39%) 4346.58 ( -2.30%)
Hmean 8192 5539.82 ( 0.00%) 3118.35 * -43.71%* 4003.50 * -27.73%* 2931.43 * -47.08%* 6069.77 ( 9.57%) 5571.91 ( 0.58%) 5143.26 ( -7.16%) 4245.56 * -23.36%*
Stddev 64 128.48 ( 0.00%) 33.63 ( 73.82%) 111.27 ( 13.39%) 64.39 ( 49.89%) 79.47 ( 38.14%) 189.98 ( -47.87%) 89.67 ( 30.20%) 135.17 ( -5.21%)
Stddev 256 191.78 ( 0.00%) 252.00 ( -31.40%) 225.33 ( -17.49%) 123.00 ( 35.86%) 183.12 ( 4.52%) 231.23 ( -20.57%) 161.13 ( 15.98%) 66.95 ( 65.09%)
Stddev 2048 463.85 ( 0.00%) 1364.71 (-194.21%) 1390.20 (-199.71%) 382.49 ( 17.54%) 1271.89 (-174.20%) 1058.81 (-128.27%) 602.49 ( -29.89%) 595.07 ( -28.29%)
Stddev 8192 1230.77 ( 0.00%) 2402.19 ( -95.18%) 567.52 ( 53.89%) 511.30 ( 58.46%) 2551.77 (-107.33%) 1065.61 ( 13.42%) 1070.69 ( 13.01%) 512.77 ( 58.34%)
As many other instances: baremetal is suffering with core scheduling.
VMs are doing reasonably good, especially when there is overcommit.
---
Ok, that's it for now... Any comment, discussion, feedback, etc, more
than welcome.
Thanks and Regards
--
Dario Faggioli, Ph.D
http://about.me/dario.faggioli
Virtualization Software Engineer
SUSE Labs, SUSE https://www.suse.com/
-------------------------------------------------------------------
<<This happens because _I_ choose it to happen!>> (Raistlin Majere)
Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)
Powered by blists - more mailing lists