That's certainly not a GPU methodology. That's a message-passing methodology that fits clusters. GPUs don't really fit that topology very well... The primary problem is that GPUs are designed to address the "data-parallelism" types of problems. That is not chess, thanks to alpha/beta's non-serial order of searching a tree.smatovic wrote:My YBWC approach on the GPU def failed now, but i found another candidate:
APHID: Asynchronous Parallel Game-Tree Search by Mark G. Brockington and Jonathan Schaeffer
--
Srdja
Possible Search Algorithms for GPUs?
Moderators: hgm, Rebel, chrisw
-
- Posts: 20943
- Joined: Mon Feb 27, 2006 7:30 pm
- Location: Birmingham, AL
Re: Possible Search Algorithms for GPUs? APHID
-
- Posts: 2645
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
Re: Possible Search Algorithms for GPUs? new gpus
...and the AMD HD 78xx series will be released in March,The 7770 and 7750 are right around the corner Smile They are more reasonably priced.
the 7850 is going to have 1280 Cores on die equipped with up to 2 GB with an 256 bit memory interface...that one sounds like an good update for me.
--
Srdja
-
- Posts: 2645
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
Re: Possible Search Algorithms for GPUs?
...i didnt get how the RBFM does his child selection....so i will try the familiar B* as next.
--
Srdja
--
Srdja