[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] use fixed point arithmetics in accumulat_influence
|
From: |
Gunnar Farneback |
|
Subject: |
Re: [gnugo-devel] use fixed point arithmetics in accumulat_influence |
|
Date: |
Fri, 29 Nov 2002 23:12:30 +0100 |
|
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
Arend wrote:
> My main motivation for this patch was to make GNU Go more suitable
> for a submission to SPECint by converting the biggest (computation
> time wise) use of floating point to integer. It turned out to be
> faster (on my Athlon K6-2 using gcc 2.95.3 -O2 -m486), and so I'd
> suggest to add this patch to CVS. (But I have no strong opinion
> here, see caveat below.)
How big speedup did you experience?
My main concern is that I find the code harder to read (and thus to
maintain) with fixed point arithmetics.
Btw, would it be meaningful to limit the fixed-point arithmetics to
accumulate_influence() and the surrounding loops?
/Gunnar