Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: questions about using popcount function to count bitboard

Author: Frank Phillips

Date: 14:06:54 01/10/03

Go up one level in this thread


On January 10, 2003 at 17:01:12, Dann Corbit wrote:

>On January 10, 2003 at 16:42:55, Frank Phillips wrote:
>[snip]
>>If you are using an Athlon, then you might wish to compare the speed of your
>>program using the non assembler and assembler versions of FirstBit, LastBit and
>>PopCount, just to check which is faster.
>>
>>As Bruce already hinted, you might also do a profile to check how significant
>>PopCount is to overall speed.
>
>It is very important that you benchmark it within the application, also.
>
>As a standalone test, with just a C version and assembly version, likely the
>assembly version will beat the pants off of the C version.  However, the
>assembly code will often defeat or confuse the optimizer.  The net effect can
>even be a slowdown when inserted into real code.

From what I said, I think we agree.  My program is faster with the non-assembler
version, which was surprising.  (PopCount is relatively insignificant for me.)

Frank



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.