[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20120105115616.GA10037@openwall.com>
Date: Thu, 5 Jan 2012 15:56:16 +0400
From: Solar Designer <solar@...nwall.com>
To: john-dev@...ts.openwall.com
Subject: Re: gcc versions
On Thu, Jan 05, 2012 at 01:41:32AM +0100, magnum wrote:
> On 01/05/2012 12:11 AM, Solar Designer wrote:
> >4.0.0 - 383K c/s, 71879 bytes (this old version of gcc generates function
> >calls for SSE2 intrinsics)
>
> For 32-bit MD4/MD5/SHA1, we have -sse2 that only use asm, and -sse2i
> that use precompiled intrinsics.
FYI, my many gcc builds are for x86-64 only, without multilib. So all
of the benchmark results I posted yesterday were for x86-64 builds.
> BTW, would you
> be able to try my testpara make targets with all those versions of gcc?
I just did - again, for x86-64 only. Here are the results.
Surprisingly, gcc 4.0.0 performed just fine - no need to disable these
intrinsics for it. The 4.4.0 vs. 4.4.6 difference is curious. I may
need to make more 4.4.x builds.
model name : Intel(R) Xeon(R) CPU E5420 @ 2.50GHz
gcc version: gcc (gcc version 4.0.0)
Best paras:
raw-MD4: 1 (25993K c/s)
crypt-MD5: 1 (15237 c/s)
raw-SHA1: 1 (9953K c/s)
--
gcc version: gcc (gcc version 4.1.0)
Best paras:
raw-MD4: 1 (24889K c/s)
crypt-MD5: 1 (15396 c/s)
raw-SHA1: 1 (9516K c/s)
--
gcc version: gcc (gcc version 4.1.2)
Best paras:
raw-MD4: 1 (24490K c/s)
crypt-MD5: 1 (15408 c/s)
raw-SHA1: 1 (9523K c/s)
--
gcc version: gcc (gcc version 4.2.0)
Best paras:
raw-MD4: 1 (24994K c/s)
crypt-MD5: 1 (15469 c/s)
raw-SHA1: 1 (9736K c/s)
--
gcc version: gcc (gcc version 4.2.4)
Best paras:
raw-MD4: 1 (25166K c/s)
crypt-MD5: 1 (15494 c/s)
raw-SHA1: 1 (9646K c/s)
--
gcc version: gcc (gcc version 4.3.0 (GCC) )
Best paras:
raw-MD4: 1 (25657K c/s)
crypt-MD5: 1 (15472 c/s)
raw-SHA1: 1 (9403K c/s)
--
gcc version: gcc (gcc version 4.3.6 (GCC) )
Best paras:
raw-MD4: 1 (25794K c/s)
crypt-MD5: 1 (15464 c/s)
raw-SHA1: 1 (9411K c/s)
--
gcc version: gcc (gcc version 4.4.0 (GCC) )
Best paras:
raw-MD4: 1 (24243K c/s)
crypt-MD5: 1 (14948 c/s)
raw-SHA1: 1 (10148K c/s)
--
gcc version: gcc (gcc version 4.4.6 (GCC) )
Best paras:
raw-MD4: 3 (35556K c/s)
crypt-MD5: 3 (22387 c/s)
raw-SHA1: 1 (10360K c/s)
--
gcc version: gcc (gcc version 4.5.0 (GCC) )
Best paras:
raw-MD4: 2 (34665K c/s)
crypt-MD5: 2 (21264 c/s)
raw-SHA1: 1 (10308K c/s)
--
gcc version: gcc (gcc version 4.5.3 (GCC) )
Best paras:
raw-MD4: 2 (34485K c/s)
crypt-MD5: 2 (21312 c/s)
raw-SHA1: 1 (10315K c/s)
--
gcc version: gcc (gcc version 4.6.0 (GCC) )
Best paras:
raw-MD4: 3 (44045K c/s)
crypt-MD5: 3 (25307 c/s)
raw-SHA1: 2 (11425K c/s)
--
gcc version: gcc (gcc version 4.6.2 (GCC) )
Best paras:
raw-MD4: 3 (43974K c/s)
crypt-MD5: 3 (25350 c/s)
raw-SHA1: 2 (11472K c/s)
Alexander
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
