From: Yi Sun <yi.y.sun@linux.intel.com>
To: "Michael Kelley (EOSG)" <Michael.H.Kelley@microsoft.com>
Cc: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"x86@kernel.org" <x86@kernel.org>,
"chao.p.peng@intel.com" <chao.p.peng@intel.com>,
"chao.gao@intel.com" <chao.gao@intel.com>,
"isaku.yamahata@intel.com" <isaku.yamahata@intel.com>,
"michael.h.kelly@microsoft.com" <michael.h.kelly@microsoft.com>,
Tianyu Lan <Tianyu.Lan@microsoft.com>,
KY Srinivasan <kys@microsoft.com>,
Haiyang Zhang <haiyangz@microsoft.com>,
Stephen Hemminger <sthemmin@microsoft.com>,
Jonathan Corbet <corbet@lwn.net>
Subject: Re: [PATCH v1 0/3] Enable PV qspinlock for Hyper-V
Date: Mon, 17 Sep 2018 22:37:54 +0800 [thread overview]
Message-ID: <20180917143754.GD11769@yi.y.sun> (raw)
In-Reply-To: <CY4PR21MB07735BAA5C00C933C4E00C9FDC180@CY4PR21MB0773.namprd21.prod.outlook.com>
On 18-09-15 23:43:12, Michael Kelley (EOSG) wrote:
> >From Yi Sun Sent: Thursday, September 13, 2018 2:13 AM
>
> > This patch adds the necessary Hyper-V specific code to allow
> > PV qspinlock work on Hyper-V.
> >
>
> Have you done any performance measurements with this
> new code, so that we know whether there is any improvement,
> or even potentially any degradation in some circumstances?
>
> Michael
>
I executed reaim in Guest (20 vcpus, 30G memory). Reaim is one of
the performance test methods used by qspinlock patch set. Results
are below. With hv_spinlock enabled, performance is better.
hv_spinlock disabled:
$ src/reaim -c data/reaim.config -f data/workfile.compute -i 16 -e 256
Num Parent Child Child Jobs per Jobs/min/ Std_dev Std_dev JTI
Forked Time SysTime UTime Minute Child Time Percent
1 1.16 0.05 1.08 5224.14 5224.14 0.00 0.00 100
17 1.40 0.75 19.22 73585.71 4328.57 0.05 4.03 95
33 2.17 1.29 36.71 92156.68 2792.63 0.12 6.13 93
49 2.84 1.59 50.50 104556.34 2133.80 0.17 6.68 93
65 3.71 2.17 68.11 106172.51 1633.42 0.24 7.14 92
81 4.63 2.87 84.00 106017.28 1308.86 0.35 8.31 91
97 5.55 3.40 101.11 105913.51 1091.89 0.32 6.27 93
113 6.28 3.90 117.70 109041.40 964.97 0.38 6.51 93
129 7.20 4.44 134.00 108575.00 841.67 0.62 9.31 90
145 8.02 4.98 150.90 109563.59 755.61 0.61 8.18 91
178 9.75 6.46 184.96 110633.85 621.54 0.62 6.78 93
211 11.91 7.65 226.10 107360.20 508.82 0.88 7.96 92
244 13.30 8.28 253.82 111175.94 455.64 0.80 6.44 93
Max Jobs per Minute 111175.94
hv_spinlock enabled:
Num Parent Child Child Jobs per Jobs/min/ Std_dev Std_dev JTI
Forked Time SysTime UTime Minute Child Time Percent
1 1.07 0.04 1.00 5663.55 5663.55 0.00 0.00 100
17 1.21 0.58 17.63 85140.50 5008.26 0.03 3.05 96
33 1.88 1.18 34.22 106372.34 3223.40 0.09 5.27 94
49 3.04 1.83 56.27 97677.63 1993.42 0.20 7.35 92
65 3.66 2.26 68.64 107622.95 1655.74 0.26 7.86 92
81 4.50 2.85 84.45 109080.00 1346.67 0.25 6.03 93
97 5.33 3.29 100.55 110285.18 1136.96 0.36 7.43 92
113 6.20 3.87 117.76 110448.39 977.42 0.39 6.87 93
129 6.96 4.42 132.93 112318.97 870.69 0.45 6.92 93
145 7.86 5.01 151.02 111793.89 770.99 0.50 6.76 93
179 9.71 6.16 185.62 111713.70 624.10 0.72 7.92 92
213 11.70 7.43 224.77 110323.08 517.95 0.73 6.69 93
247 13.45 8.48 259.19 111287.73 450.56 0.73 5.78 94
Max Jobs per Minute 112318.97
prev parent reply other threads:[~2018-09-17 14:42 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-13 9:13 [PATCH v1 0/3] Enable PV qspinlock for Hyper-V Yi Sun
2018-09-13 9:13 ` [PATCH v1 1/3] X86/Hyper-V: Add Guest IDLE MSR support Yi Sun
2018-09-13 9:13 ` [PATCH v1 2/3] locking/pvqspinlock, hv: Enable PV qspinlock for Hyper-V Yi Sun
2018-09-13 11:06 ` Thomas Gleixner
2018-09-14 8:53 ` Yi Sun
2018-09-13 11:24 ` Thomas Gleixner
2018-09-14 9:04 ` Yi Sun
2018-09-14 10:42 ` Thomas Gleixner
2018-09-17 12:54 ` Yi Sun
2018-09-20 1:56 ` Yi Sun
2018-09-13 16:16 ` kbuild test robot
2018-09-13 16:42 ` kbuild test robot
2018-09-13 9:13 ` [PATCH v1 3/3] hv: add description for hv_nopvspin Yi Sun
2018-09-15 23:43 ` [PATCH v1 0/3] Enable PV qspinlock for Hyper-V Michael Kelley (EOSG)
2018-09-17 14:37 ` Yi Sun [this message]
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=20180917143754.GD11769@yi.y.sun \
--to=yi.y.sun@linux.intel.com \
--cc=Michael.H.Kelley@microsoft.com \
--cc=Tianyu.Lan@microsoft.com \
--cc=chao.gao@intel.com \
--cc=chao.p.peng@intel.com \
--cc=corbet@lwn.net \
--cc=haiyangz@microsoft.com \
--cc=isaku.yamahata@intel.com \
--cc=kys@microsoft.com \
--cc=linux-kernel@vger.kernel.org \
--cc=michael.h.kelly@microsoft.com \
--cc=sthemmin@microsoft.com \
--cc=x86@kernel.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;
as well as URLs for NNTP newsgroup(s).