Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Profiled Crafty 18.13

Author: Jeremiah Penery

Date: 17:06:19 04/17/02

Go up one level in this thread


On April 17, 2002 at 13:07:34, Slater Wold wrote:

>Most expensive:
>
>Evaluate:  537 hits per ms.  (1)
>GenerateCaptures: 503 hits per ms.  (3)
>Search: 185 hits per ms.  (5)
>ClearHashTable: 0.0031 hits per ms.  (16)
>
>Most called:
>
>PopCnt: 4401 hits per ms.  (10)
>FirstOnes: 4387 hits per ms.  (2)
>LastOnes: 4273 hits per ms.  (7)
>MakeMove: 4219 hits per ms.  (14)
>EvaluateMaterial: 4159 hits per ms.  (15)
>
>
>629 seconds to clear the hash table, twice (96M + 48M).

It shouldn't take more than 2 seconds to clear the hash table...

>Rank of CPU usage is in the ().

A couple of things:

It looked like you were not using the x86 assembly (firstone/lastone/popcnt in
boolean.c instead of vcinline.h).  The assembly functions are a ton faster, why
not use them?

Why does it show
16267.356  33.6    16267.356  33.6        6 _ReadInput (utility.obj)
as the first line in the profile?  Is there something that made this function
use a ton of time?

Lastly, how do I do a profile run in MSVC, if you know? :)  I've added auxiliary
functions to the evaluation, and I want to see which one(s) are the most
expensive.

Thanks,
Jeremiah



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.