Search found 25 matches

by johnhamlen
Mon Jul 02, 2012 11:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: link to paper on solving games on the GPU
Replies: 3
Views: 1105

Re: link to paper on solving games on the GPU

Interesting, helpful, and well-timed. Thanks for posting Rein, and congratulations to Damian for finishing the work.
by johnhamlen
Mon Jun 11, 2012 11:41 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there such a thing as branchless move generation?
Replies: 55
Views: 17386

Re: Is there such a thing as branchless move generation?

Sorry Daniel, I didn't mean to annoy/upset/frustrate you in any way, but clearly I have and I apologise :(. I sincerely appreciated all your points and have learned a lot. Good night and have a great week.

John
by johnhamlen
Mon Jun 11, 2012 11:14 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Choosing a GPU platform: AMD and Nvidia
Replies: 16
Views: 5274

Re: Choosing a GPU platform: AMD and Nvidia

I am casually reading about the kepler architecture now. Despite the huge increase in number of cores per SMX, number of registers barely increased while the shared mem & cache are kept the same at 64kb now serving even more threads :( Well it seems it is not going to help us chess programmers a lo...
by johnhamlen
Mon Jun 11, 2012 10:35 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there such a thing as branchless move generation?
Replies: 55
Views: 17386

Re: Zeta CL uses Magic Bitboards

Hi John, maybe you want to take a look at "Zeta CL" http://zeta-chess.blogspot.de/ Code is published under GPL: https://github.com/smatovic/Zeta/ Zeta uses QuadBitboards (thx to Gerd) for Board Presentation and a Magic-Bitboard Move Generator (parts ported from Stockfish). I tried an 0x88 move gene...
by johnhamlen
Mon Jun 11, 2012 10:12 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there such a thing as branchless move generation?
Replies: 55
Views: 17386

Re: Is there such a thing as branchless move generation?

Well that is not a piece list. LOL. Okay then, it's a list of squares that the pieces are on :wink:. Wow, it's a tough crowd in tonight! :D.... You should be able to loop over pieces of certain kind without testing any flag. If that was the case, why not do what TSCP does with only board[64]? The c...
by johnhamlen
Mon Jun 11, 2012 1:06 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Choosing a GPU platform: AMD and Nvidia
Replies: 16
Views: 5274

Re: Choosing a GPU platform: AMD and Nvidia

Well I am not that familiar with the kepler as that is brand new. It seems there are far more number of cores per SM now after some innovations. For the fermi it was 32 now it is 192. Anyway you have to remeber that an SM runs a lot more threads than the number of cores. So for example the fermi ca...
by johnhamlen
Mon Jun 11, 2012 12:43 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there such a thing as branchless move generation?
Replies: 55
Views: 17386

Re: Is there such a thing as branchless move generation?

You still need a pointer to the next element if it is to be a linked list. Agreed, but I didn't explain myself very well. The idea was to not have any sort of linked list, just an array of 32 entries, one for each possible piece on the board. Before you mentioned issues with memory alignment I was ...
by johnhamlen
Sun Jun 10, 2012 11:25 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Choosing a GPU platform: AMD and Nvidia
Replies: 16
Views: 5274

Re: Choosing a GPU platform: AMD and Nvidia

No. Shared memory (or. local memory as AMD calls it) is allocated per multi-processor not for each cuda core. Each mp consists of a group of 8 cuda cores, and the latest one using fermi architecture have 32 cuda cores per mp. Infact the 48kb figure is for the fermi, so you get very little per core....
by johnhamlen
Sun Jun 10, 2012 10:58 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there such a thing as branchless move generation?
Replies: 55
Views: 17386

Re: Is there such a thing as branchless move generation?

Last time I checked NVIDIA tesla's have 64kb that you can divide into 48kb + 16kb shared memory and local cache. If you just want the device to do most of the caching of global mem for you, use 48kb for that otherwise 48kb of shared memory per multi-processor is what you get. Cheers Daniel. Now all...
by johnhamlen
Sun Jun 10, 2012 10:14 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there such a thing as branchless move generation?
Replies: 55
Views: 17386

Re: Is there such a thing as branchless move generation?

Your calculation for piece list is totally wrong, at least according to what I am using now. The piece list entries with LIST { int sq, List* prev,List* next } requires far more than a thousand bytes. You need to declare an LIST[128] too... Agreed. Totally wrong if I was calculating for a doubly li...