Author: Stuart Cracraft
Date: 21:00:00 08/04/04
Go up one level in this thread
On August 04, 2004 at 23:38:46, Stuart Cracraft wrote:
>Nothing technical here that anyone can really chew
>on and I don't expect any huge insights from myself
>or anyone; but I have had nothing but bitter frustration
>after trying to implement something ostensibly simple:
>avoiding a full move generation after a successful
>has lookup. My version with that feature is still
>8.5% slower than the version that does a full move gen
>and looks for the hash move afterwards.
>
>Nothing like banging one's head against a concrete wall
>for a week.
This bulletin board is some kind of weird Rogerian Reflective
counselor's couch! So I worked on this problem for 1+ week
and was ready to give up, write a simple post, and 5 minutes
later find it.
New program is:
19.5% faster than previous non-optimal attempt with same
movegen-after-hashmove
and
12.5% faster than the prior attempt with movegen-before-hashmove
The code is certainly uglier with special exceptions and a three-way
flag for various conditions of the hash move, but when all was said
and done, the pizza man delivered.
Now on to SEE and I see there is a helpful thread already out there.
Stuart
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.