From: mark gross <mgross@linux.intel.com>
To: speck@linutronix.de
Subject: [MODERATED] Re: [PATCH 0/4] walnut 0
Date: Fri, 1 Mar 2019 10:59:44 -0800 [thread overview]
Message-ID: <20190301185944.GA21593@mgross-MOBL.amr.corp.intel.com> (raw)
In-Reply-To: <20190219155807.878855982@infradead.org>
[-- Attachment #1: Type: text/plain, Size: 812 bytes --]
On Tue, Feb 19, 2019 at 04:58:07PM +0100, speck for Peter Zijlstra wrote:
> From: Peter Zijlstra <peterz@infradead.org>
> Subject: [PATCH 0/4] walnut
>
> nicely split
Peter, We have been testing your patch set named after a nut posted Feb 19.
We tested it on bare metal and it looks good. There was some worry that a KVM
guest would have issues with multiplexing if the host had your patch and
allow_tsx_abort=0.
What we saw away the 4th counter in the guest showed zero. Which IMO seems ok.
Attached are the tests that Nelson ran for both host and kvm guest. The
behavior between host and guest behavior with 4 counters when allow_tsx_abort=0
looks minor to me.
We are eager to see a more final version of this patchset so we can be ready
for the public disclosure coming up March 12.
Thanks! --mark
[-- Attachment #2: patchedhost.txt --]
[-- Type: text/plain, Size: 2859 bytes --]
/* ALLOW_TSX_FORCE_ABORT = 0 */
// Four Counters
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches,branch-misses}' --no-merge ./rtm-success-long
rtm success 100.00%, aborted 155
Performance counter stats for './rtm-success-long':
154 cs
<not counted> cache-misses
<not counted> cache-references
<not counted> branches
<not supported> branch-misses
6.749371782 seconds time elapsed
4.458202000 seconds user
0.000000000 seconds sys
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 193
// Three Counters
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches}' --no-merge ./rtm-success-long
rtm success 100.00%, aborted 171
Performance counter stats for './rtm-success-long':
155 cs
5,332 cache-misses # 16.190 % of all cache refs
32,934 cache-references
201,012,996 branches
6.808987753 seconds time elapsed
4.466809000 seconds user
0.000000000 seconds sys
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 166
/* ALLOW_TSX_FORCE_ABORT = 1 */
//Four Counters
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches,branch-misses}' --no-merge ./rtm-success-long
rtm success 0.00%, aborted 100000000
Performance counter stats for './rtm-success-long':
145 cs
42,505 cache-misses # 48.776 % of all cache refs
87,144 cache-references
301,075,199 branches
100,040,329 branch-misses # 33.23% of all branches
6.496993416 seconds time elapsed
4.347726000 seconds user
0.000000000 seconds sys
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 589
//Three Counters
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches}' --no-merge ./rtm-success-long
rtm success 100.00%, aborted 162
Performance counter stats for './rtm-success-long':
145 cs
5,488 cache-misses # 17.551 % of all cache refs
31,268 cache-references
201,016,636 branches
6.593972485 seconds time elapsed
4.462666000 seconds user
0.000000000 seconds sys
root@jf1-otc-3AR3-33:/home/labuser/ocean/tools/perf# taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 180
[-- Attachment #3: patchedkvm_guest.txt --]
[-- Type: text/plain, Size: 2992 bytes --]
/* ALLOW_TSX_FORCE_ABORT = 0 */
// Four Counters
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches,branch-misses}' --no-merge ./rtm-success-long
rtm success 100.00%, aborted 919
Performance counter stats for './rtm-success-long':
0 cs:u
3,318 cache-misses:u # 44.753 % of all cache refs
7,414 cache-references:u
200,038,222 branches:u
0 branch-misses:u # 0.00% of all branches
16.023427318 seconds time elapsed
6.446820000 seconds user
2.420087000 seconds sys
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 701
// Three Counters
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches}' --no-merge ./rtm-success-long
rtm success 100.00%, aborted 830
Performance counter stats for './rtm-success-long':
0 cs:u
3,311 cache-misses:u # 47.354 % of all cache refs
6,992 cache-references:u
200,038,067 branches:u
13.679651620 seconds time elapsed
6.123687000 seconds user
1.926116000 seconds sys
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 704
/* ALLOW_TSX_FORCE_ABORT = 1 */
// Four Counters
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches,branch-misses}' --no-merge ./rtm-success-long
rtm success 0.00%, aborted 100000000
Performance counter stats for './rtm-success-long':
0 cs:u
2,742 cache-misses:u # 29.685 % of all cache refs
9,237 cache-references:u
300,037,068 branches:u
100,003,687 branch-misses:u # 33.33% of all branches
17.318943108 seconds time elapsed
6.415269000 seconds user
2.753311000 seconds sys
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 1501
//Three Counters
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./perf stat -e cs,'{cache-misses,cache-references,branches}' --no-merge ./rtm-success-long
rtm success 100.00%, aborted 772
Performance counter stats for './rtm-success-long':
0 cs:u
2,892 cache-misses:u # 40.228 % of all cache refs
7,189 cache-references:u
200,038,116 branches:u
17.638868052 seconds time elapsed
6.710811000 seconds user
2.761211000 seconds sys
kvm-virtual@kvmvirtual-Standard-PC-i440FX-PIIX-1996:~$ taskset -c 0 ./rtm-success-long
rtm success 100.00%, aborted 830
prev parent reply other threads:[~2019-03-01 18:59 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-19 15:58 [MODERATED] [PATCH 0/4] walnut 0 Peter Zijlstra
2019-02-19 15:58 ` [MODERATED] [PATCH 1/4] walnut 1 Peter Zijlstra
2019-02-19 15:58 ` [MODERATED] [PATCH 2/4] walnut 2 Peter Zijlstra
2019-02-19 15:58 ` [MODERATED] [PATCH 3/4] walnut 3 Peter Zijlstra
2019-02-19 15:58 ` [MODERATED] [PATCH 4/4] walnut 4 Peter Zijlstra
2019-02-19 16:10 ` [MODERATED] " Peter Zijlstra
2019-02-20 22:31 ` Nelson D'Souza
2019-02-22 22:58 ` [MODERATED] Fwd: " Nelson D'Souza
2019-02-22 23:19 ` Nelson D'Souza
2019-02-19 18:49 ` [MODERATED] Linus Torvalds
2019-02-20 14:37 ` Peter Zijlstra
2019-03-07 13:56 ` [MODERATED] Re: [PATCH 4/4] walnut 4 Greg KH
2019-03-07 14:34 ` Peter Zijlstra
2019-03-07 15:09 ` mark gross
2019-02-20 18:10 ` [MODERATED] Re: [PATCH 0/4] walnut 0 mark gross
2019-02-22 19:20 ` Nelson D'Souza
2019-03-01 18:59 ` mark gross [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=20190301185944.GA21593@mgross-MOBL.amr.corp.intel.com \
--to=mgross@linux.intel.com \
--cc=speck@linutronix.de \
/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).