|It's 1997. Do you know where your scheduler is? email@example.com (D. J. Bernstein) (1997-12-19)|
|Re: It's 1997. Do you know where your scheduler is? firstname.lastname@example.org (David Chase) (1997-12-23)|
|Re: It's 1997. Do you know where your scheduler is? email@example.com (D. J. Bernstein) (1997-12-29)|
|Re: It's 1997. Do you know where your scheduler is? firstname.lastname@example.org (David Greene) (1998-01-04)|
|From:||David Greene <email@example.com>|
|Date:||4 Jan 1998 20:51:20 -0500|
|Organization:||University of Michigan EECS|
|Keywords:||architecture, performance, optimize|
D. J. Bernstein <firstname.lastname@example.org> wrote:
: > which release of gcc,
: It really doesn't matter. Here are some Pentium cycle counts for the
: same hand-scheduled 256-point complex FFT code:
: 23085 (best) gcc 22.214.171.124 -O1 -fo-f-p
: 41913 gcc 126.96.36.199 -O6 -fo-f-p
: 47258 egcs 1.00 -O6 -fo-f-p -mpentiumpro
: 56383 egcs 1.00 -O6 -fo-f-p -mpentium
: 56860 (worst) egcs 1.00 -O6 -fo-f-p
Has anyone tried egcs? It apparently contains a Pentium scheduler.
I don't know whether it does PPro.
: The Pentium Pro, as you noted, is a very different chip. Scheduling
: code badly for the Pentium Pro would take quite a bit of effort.
Unless you ignore the 4-1-1 decode, which is (IIRC) the main
bottleneck (once stuff is in from the cache/memory, that is...).
Return to the
Search the comp.compilers archives again.