netdev - Re: [PATCH] x86: Run checksumming in parallel accross multiple alu's

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20131101092148.GB27063@gmail.com>
Date:	Fri, 1 Nov 2013 10:21:48 +0100
From:	Ingo Molnar <mingo@...nel.org>
To:	Neil Horman <nhorman@...driver.com>
Cc:	Doug Ledford <dledford@...hat.com>,
	Eric Dumazet <eric.dumazet@...il.com>,
	linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
	David Laight <David.Laight@...LAB.COM>
Subject: Re: [PATCH] x86: Run checksumming in parallel accross multiple alu's


* Neil Horman <nhorman@...driver.com> wrote:

> Prefetch and simluated adcx/adox from above:
>  Performance counter stats for './test.sh' (20 runs):
> 
>         35,704,331 L1-dcache-load-misses                                         ( +-  0.07% ) [75.00%]
>                  0 L1-dcache-prefetches                                         [75.00%]
>     19,751,409,264 cycles                    #    0.000 GHz                      ( +-  0.59% ) [75.00%]
>         34,850,056 branch-misses                                                 ( +-  1.29% ) [75.00%]
> 
>        7.768602160 seconds time elapsed                                          ( +-  1.38% )

btw., you might also want to try measuring only the basics:

   -e cycles -e instructions -e branches -e branch-misses

that should give you 100% in the last column and should also allow 
you to double check whether all the PMU counts are correct: is it 
the expected number of instructions, expected number of branches, 
expected number of branch-misses, etc.

Then you can remove branch stats and add just L1-dcache stats - and 
still be 100% covered:

   -e cycles -e instructions -e L1-dcache-loads -e L1-dcache-load-misses

etc.

Just so that you can trust what the PMU tells you. Prefetch counts 
are sometimes off, they might include speculative activities, etc.

Thanks,

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html