Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: piece-lists versus bitboards on 64-bit-systems

Author: Vasik Rajlich

Date: 03:01:09 03/13/05

Go up one level in this thread


On March 13, 2005 at 04:18:59, F. Reul wrote:

>Hello,
>a few weeks ago, I read an article in "ct-magazin" about the 64-bit-world. The
>benchmark-test "186.crafty" promises 20 percent speed-increase with the new
>64-bit-compilers of Intel and Microsoft with Windows x64 on a 64-bit-system.
>
>My question, can some programmers confirm these results, are there any articles
>online to read about this, and is a speed-increase of bitboard-engines more
>than 30 percent possible or not?
>
>Can the "microsoft-whidbey-compiler" optimize on 64-bit-systems with
>"feedback-protocol", or is the "intel 8.1-compiler" better for bitboard-engines
>?
>
>What´s about the speed-increase of classic-32-bit engines (0x88 or similar
>architectures) ?
>
>I would be very happy, about a discussion :-))
>
>Fritz Reul (List)

Just about every engine runs significantly faster in 64 bits. It's been a while
since I measured, but Rybka (rotated bitboards) does around 1100 Knps on a
64-bit Amd 3400+, 750 Knps on the same machine in 32-bit mode, and 650 Knps or
so on a 3.2 GHz pentium. My 64-bit speedup was in the neighborhood of 50%.

Vas



This page took 0 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.