From: Vince Weaver <vince@deater.net>
To: Ingo Molnar <mingo@elte.hu>
Cc: linux-kernel@vger.kernel.org,
Thomas Gleixner <tglx@linutronix.de>,
Andrew Morton <akpm@linux-foundation.org>,
Stephane Eranian <eranian@googlemail.com>,
Eric Dumazet <dada1@cosmosbay.com>,
Robert Richter <robert.richter@amd.com>,
Arjan van de Ven <arjan@infradead.org>,
Peter Anvin <hpa@zytor.com>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
Paul Mackerras <paulus@samba.org>,
"David S. Miller" <davem@davemloft.net>,
perfctr-devel@lists.sourceforge.net
Subject: Re: [patch] Performance Counters for Linux, v4
Date: Tue, 16 Dec 2008 12:38:59 -0500 (EST) [thread overview]
Message-ID: <Pine.LNX.4.64.0812161232170.6249@pianoman.cluster.toy> (raw)
In-Reply-To: <20081214212829.GA9435@elte.hu>
I'm trying to evaluate this new proposal for the kind of workloads I use
performance counters for, and even the simplest tests don't work.
I'm trying to do a simple aggragate count for some benchmarks here using
timec and I'm getting poor results.
Are any of the problems I'm reporting going to be fixed?
In any case, I was testing aggregate counts on a longer running benchmark,
this time equake from the spec2k benchmark suite, still on the q6600.
If I only count retired instructions, I get consistent results:
timec -e 1
119175255369 instructions (events)
119175255561 instructions (events)
119175255383 instructions (events)
however the minute I add another count, say cycles so I can calculate
CPI/IPC the results for instructions are suddenly off by 33%.
Needless to say, perfmon can handle reading both cycles and instructions
at the same time.
timec -e 0, -e 1
91758816320 cycles (events)
79428247907 instructions (events)
91849140396 cycles (events)
79449560742 instructions (events)
It gets worse when trying to look at cache statistics:
timec -e 1 -e 2 -e 3
59611457943 instructions (events)
1872499771 cache references (events)
97471971 cache misses (events)
59601907232 instructions (events)
1871766376 cache references (events)
97435199 cache misses (events)
and so on
timec -e1 -e2 -e3 -e4
47671703285 instructions (events)
1498246999 cache references (events)
77838085 cache misses (events)
3394839360 branches (events)
47666131604 instructions (events)
1497069685 cache references (events)
78065325 cache misses (events)
3393244879 branches (events)
So apparently this performance counter infrastructure will always be
useless for trying to get plain aggregate counts? It's the simplest case
to get right, so it makes me wonder about the design of the rest of the
infrastructure.
Vince
next prev parent reply other threads:[~2008-12-16 17:34 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-12-14 21:28 [patch] Performance Counters for Linux, v4 Ingo Molnar
2008-12-15 11:59 ` Paul Mackerras
2008-12-15 12:11 ` Paul Mackerras
2008-12-16 14:22 ` Peter Zijlstra
2008-12-16 23:06 ` Paul Mackerras
2008-12-16 23:51 ` Ingo Molnar
2008-12-17 1:55 ` Andi Kleen
2009-01-16 18:01 ` Corey Ashford
2009-01-16 22:14 ` Maynard Johnson
2009-01-16 23:11 ` Ingo Molnar
2009-01-17 1:26 ` Paul Mackerras
2009-01-17 9:53 ` Andi Kleen
2008-12-17 2:23 ` [Perfctr-devel] " Dan Terpstra
2008-12-17 7:34 ` stephane eranian
2008-12-15 17:44 ` Vince Weaver
2008-12-15 21:07 ` Vince Weaver
2008-12-15 22:13 ` Paul Mackerras
2008-12-15 21:42 ` Paul Mackerras
2008-12-15 22:03 ` stephane eranian
2008-12-16 14:42 ` Peter Zijlstra
2008-12-16 16:55 ` Vince Weaver
2008-12-16 21:52 ` Paul Mackerras
2008-12-16 12:22 ` Pavel Machek
2008-12-16 12:50 ` Ingo Molnar
2008-12-16 12:57 ` Pavel Machek
2008-12-16 13:03 ` Ingo Molnar
2008-12-16 13:13 ` Arjan van de Ven
2008-12-16 20:04 ` Pavel Machek
2008-12-16 14:45 ` Peter Zijlstra
2008-12-16 15:46 ` [Perfctr-devel] " Martin Cracauer
2008-12-16 17:38 ` Vince Weaver [this message]
2008-12-16 19:47 ` Corey Ashford
2008-12-16 20:55 ` Vince Weaver
2008-12-16 19:56 ` [Perfctr-devel] " William Cohen
2008-12-17 1:51 ` Andi Kleen
2008-12-17 1:56 ` Samuel Thibault
[not found] ` <d484eb1f0812162357n7a851f2fncc0abaae9bd293a4@mail.gmail.com>
2008-12-17 9:18 ` Andi Kleen
[not found] ` <d484eb1f0812170611s597e014cl5fb98d6dd81afd49@mail.gmail.com>
2008-12-17 15:06 ` Andi Kleen
2008-12-17 16:00 ` William Cohen
2008-12-17 20:53 ` Corey Ashford
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=Pine.LNX.4.64.0812161232170.6249@pianoman.cluster.toy \
--to=vince@deater.net \
--cc=a.p.zijlstra@chello.nl \
--cc=akpm@linux-foundation.org \
--cc=arjan@infradead.org \
--cc=dada1@cosmosbay.com \
--cc=davem@davemloft.net \
--cc=eranian@googlemail.com \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=paulus@samba.org \
--cc=perfctr-devel@lists.sourceforge.net \
--cc=robert.richter@amd.com \
--cc=tglx@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.