From: arun@sharma-home.net
To: Ingo Molnar <mingo@elte.hu>
Cc: Stephane Eranian <eranian@google.com>,
Arnaldo Carvalho de Melo <acme@infradead.org>,
linux-kernel@vger.kernel.org, Andi Kleen <ak@linux.intel.com>,
Peter Zijlstra <peterz@infradead.org>,
Lin Ming <ming.m.lin@intel.com>,
Arnaldo Carvalho de Melo <acme@redhat.com>,
Thomas Gleixner <tglx@linutronix.de>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
eranian@gmail.com, Arun Sharma <asharma@fb.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [generalized cache events] Re: [PATCH 1/1] perf tools: Add missing user space support for config1/config2
Date: Fri, 22 Apr 2011 09:50:07 -0700 [thread overview]
Message-ID: <20110422165007.GA18401@vps.sharma-home.net> (raw)
In-Reply-To: <20110422105211.GB1948@elte.hu>
On Fri, Apr 22, 2011 at 12:52:11PM +0200, Ingo Molnar wrote:
>
> Using the generalized cache events i can run:
>
> $ perf stat --repeat 10 -e cycles:u -e instructions:u -e l1-dcache-loads:u -e l1-dcache-load-misses:u ./array
>
> Performance counter stats for './array' (10 runs):
>
> 6,719,130 cycles:u ( +- 0.662% )
> 5,084,792 instructions:u # 0.757 IPC ( +- 0.000% )
> 1,037,032 l1-dcache-loads:u ( +- 0.009% )
> 1,003,604 l1-dcache-load-misses:u ( +- 0.003% )
>
> 0.003802098 seconds time elapsed ( +- 13.395% )
>
> I consider that this is 'bad', because for almost every dcache-load there's a
> dcache-miss - a 99% L1 cache miss rate!
One could argue that all you need is cycles and instructions. If there is an
expensive load, you'll see that the load instruction takes many cycles and
you can infer that it's a cache miss.
Questions app developers typically ask me:
* If I fix all my top 5 L3 misses how much faster will my app go?
* Am I bottlenecked on memory bandwidth?
* I have 4 L3 misses every 1000 instructions and 15 branch mispredicts per
1000 instructions. Which one should I focus on?
It's hard to answer some of these without access to all events.
While your approach of having generic events for commonly used counters
might be useful for some use cases, I don't see why exposing all vendor
defined events is harmful.
A clear statement on the last point would be helpful.
-Arun
next prev parent reply other threads:[~2011-04-22 16:59 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-04-22 8:47 Stephane Eranian
2011-04-22 9:23 ` Ingo Molnar
2011-04-22 9:41 ` Stephane Eranian
2011-04-22 10:52 ` [generalized cache events] " Ingo Molnar
2011-04-22 12:04 ` Stephane Eranian
2011-04-22 13:18 ` Ingo Molnar
2011-04-22 20:31 ` Stephane Eranian
2011-04-22 20:47 ` Ingo Molnar
2011-04-23 12:13 ` Stephane Eranian
2011-04-23 12:49 ` Ingo Molnar
2011-04-22 21:03 ` Ingo Molnar
2011-04-23 12:27 ` Stephane Eranian
2011-04-22 16:51 ` Andi Kleen
2011-04-22 19:57 ` Ingo Molnar
2011-04-26 9:25 ` Peter Zijlstra
2011-04-22 16:50 ` arun [this message]
2011-04-22 17:00 ` Andi Kleen
2011-04-22 20:30 ` Ingo Molnar
2011-04-22 20:32 ` Ingo Molnar
2011-04-23 0:03 ` Andi Kleen
2011-04-23 7:50 ` Peter Zijlstra
2011-04-23 12:06 ` Stephane Eranian
2011-04-23 12:36 ` Ingo Molnar
2011-04-23 13:16 ` Peter Zijlstra
2011-04-25 18:48 ` Stephane Eranian
2011-04-25 19:40 ` Andi Kleen
2011-04-25 19:55 ` Ingo Molnar
2011-04-24 2:15 ` Andi Kleen
2011-04-24 2:19 ` Andi Kleen
2011-04-25 17:41 ` Ingo Molnar
2011-04-25 18:00 ` Dehao Chen
[not found] ` <BANLkTiks31-pMJe4zCKrppsrA1d6KanJFA@mail.gmail.com>
2011-04-25 18:05 ` Ingo Molnar
2011-04-25 18:39 ` Stephane Eranian
2011-04-25 19:45 ` Ingo Molnar
2011-04-23 8:02 ` Ingo Molnar
2011-04-23 20:14 ` [PATCH] perf events: Add stalled cycles generic event - PERF_COUNT_HW_STALLED_CYCLES Ingo Molnar
2011-04-24 6:16 ` Arun Sharma
2011-04-25 17:37 ` Ingo Molnar
2011-04-26 9:25 ` Peter Zijlstra
2011-04-26 14:00 ` Ingo Molnar
2011-04-27 11:11 ` Ingo Molnar
2011-04-27 14:47 ` Arun Sharma
2011-04-27 15:48 ` Ingo Molnar
2011-04-27 16:27 ` Ingo Molnar
2011-04-27 19:05 ` Arun Sharma
2011-04-27 19:03 ` Arun Sharma
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=20110422165007.GA18401@vps.sharma-home.net \
--to=arun@sharma-home.net \
--cc=a.p.zijlstra@chello.nl \
--cc=acme@infradead.org \
--cc=acme@redhat.com \
--cc=ak@linux.intel.com \
--cc=akpm@linux-foundation.org \
--cc=asharma@fb.com \
--cc=eranian@gmail.com \
--cc=eranian@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=ming.m.lin@intel.com \
--cc=mingo@elte.hu \
--cc=peterz@infradead.org \
--cc=tglx@linutronix.de \
--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
all inboxes | Powered by JetHome®