Re: More performance tests of different AES implementations

From: Brian Gladman (fake_at_nowhere.org)
Date: 07/08/03 

Date: Tue, 8 Jul 2003 22:55:55 +0100

Mok-Kong Shen wrote:
> Christophe Devine wrote:
>>
> {snip]
>> Some fresh results:
> [snip]
>
> It's very surprising in my view that on Pentium 4 with
> gcc your code for enc/dec is better than Gladman's,
> while with VC++ the opposite is the case, and that that
> relationship is quite harsh. I wonder why the quality of
> compiler could be so decisive. (There must be some
> constructs in the codes that are presumably in a sense
> very 'sensitive' for the two compilers.)

This is no surprise since my version is optimised for VC++ while
Christophe's is optimised for GCC.

Most of the fast versions actually end up with the same code structure
although they get there in very different ways - which means that, in
respect of encryption speeds, this test is almost entirely a compiler
comparison.

It would be interesting to see how the BCC, the Open Watcom and the Digital
MARS compilers perform on this test.

For key scheduling, however, this exercise does test the AES code since
different authors are using very different strategies for this.

   Brian Gladman

Relevant Pages

  • Re: floating point arithmetic
    ... I don't find "dodgy" results obtained by options for targets of more than 8 years ago so surprising either. ... gcc has such a "surprising" default only for i386. ... I'm equally inclined to complain to my employer's IT for locking my laptop down to 32-bit Windows last week as to complain to c.l.c about gcc defaults. ... If you want to discuss why a certain compiler has obsolete defaults, if it is a currently maintained compiler, there should be a more suitable forum. ...
    (comp.lang.c)
  • Re: OOPS: 2.6.16-rc6 cpufreq_conservative
    ... In the version of gcc you've tested. ... The end result may be _surprising_, the same way Duff's device is ... support Duff's device is not a C compiler. ... git grep '({.*return' ...
    (Linux-Kernel)
  • Re: floating point arithmetic
    ... gcc has such a "surprising" default only for i386. ... I don't find either the gcc default nor the resulting ... 32-bit Windows last week as to complain to c.l.c about gcc defaults. ... If you want to discuss why a certain compiler has obsolete defaults, ...
    (comp.lang.c)
  • Re: More performance tests of different AES implementations
    ... gcc your code for enc/dec is better than Gladman's, ... compiler could be so decisive. ...
    (sci.crypt)
  • Re: Whats the story with the "end of XP"?
    ... Support gives them access to the larger body ... The same compiler suite may be "best" in both ... In the case of ByteCraft compilers, you are truly expert in your field, you work closely and rapidly with customers if there are any issues, and you have top class testing and quality control. ... understanding the inner workings of a relatively simple compiler like GCC ...
    (comp.arch.embedded)