Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 64 bits

Author: Tom Kerrigan

Date: 12:50:42 06/20/02

Go up one level in this thread


On June 20, 2002 at 14:25:17, Eugene Nalimov wrote:

>As you pointed out that not necessarily means that 64-bit CPUs are good for
>Crafty because of the 64-bit data path, but there are good chances that 64-bit
>data path helps. The question is how much. Unfortunately, experiment you
>suggested is flawed, as I wrote in one of my recent postings, and I don't see
>another quick way to answer the question. My "gut feeling" is ~10%.

Do you really think it's flawed? Compilers can't mess up logical operations
(which would be the majority of operations done on bitboards) that badly, can
they?

Other possibilities to determine the speedup:

1) Rework gcc to produce 32-bit code for a 64-bit chip. Do you know if this
would be hard? I can't imagine it'd be THAT hard, but probably harder than
making the bitboard class.

2) Wait for Hammers and see how fast Crafty runs in each mode, although that's
not ideal because x86-64 mode also gives you an extra 8 registers.

-Tom



This page took 0.01 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.