Search found 70 matches

by ankan
Sun Nov 06, 2016 5:10 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Pigeon is now running on the GPU
Replies: 13
Views: 3246

Re: Pigeon is now running on the GPU

what search algorithm are you using?
You mentioned negaMax - is that with or without alpha-beta pruning?
by ankan
Thu Oct 20, 2016 5:17 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Jetson GPU architecture
Replies: 5
Views: 1468

Re: Jetson GPU architecture

If the same technological innovation makes its way to the big GPU systems, then we would see a marvelous increase in compute power from truly heterogeneous CPU/GPU compute engines. Actually there are slightly bigger systems that already have similar technology - game consoles. For example PS4 has a...
by ankan
Thu Oct 20, 2016 4:34 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Jetson GPU architecture
Replies: 5
Views: 1468

Re: Jetson GPU architecture

I haven't personally used any of these embedded systems with unified CPU and GPU memory, but as Srdja pointed out they are very low-end/small machines compared to discrete PCIE GPUs (like at least an order of magnitude slower). In some situations where you need to pass a lot of data or need tightly ...
by ankan
Tue Sep 13, 2016 6:30 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: yet another attempt on Perft(14)
Replies: 33
Views: 10462

Re: yet another attempt on Perft(14)

I don't know if this implies that there were more than one bit flip in the first run, or if the same flipped bit stayed around in the hash table long enough to affect several sub-trees. Impossible to say. I think there is a good probability of same flipped bit staying in the hash table, but we can'...
by ankan
Fri Sep 09, 2016 8:38 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: yet another attempt on Perft(14)
Replies: 33
Views: 10462

Re: yet another attempt on Perft(14)

Here is the result from the second run (actually two partial runs as the run got interrupted): ..a2-a3 1707945110805813310 ..a2-a4 2493115507450124588 ..b1-a3 1988096314966752669 ..b1-c3 2638654908094170585 ..b2-b3 2286406377370528139 ..b2-b4 2297640117530334747 ..c2-c3 2666334464745721380 ..c2-c4 3...
by ankan
Wed Aug 31, 2016 3:45 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Perft(14) Weekly Status Reports for 2016
Replies: 75
Views: 21483

Re: Work unit 017

For [d]rnbqkbnr/1pQ1ppp1/8/3p4/8/4P3/PPPP1PPP/RNB1KBNR b KQkq - 0 4[/d] using my CPU perft program: Perft 7: 92374081671, Time taken: 44.0251 seconds, nps: 2098213359 Something looks wrong. I computed these results on a different (older) machine with a single GPU (which was overclocked). Maybe it's ...
by ankan
Tue Aug 30, 2016 6:41 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Perft(14) Weekly Status Reports for 2016
Replies: 75
Views: 21483

Re: Perft(14) Weekly Status 2016-08-28

I have computed the first 100 work units using my updated program (with 128 bit hashes). Here are the results:
https://drive.google.com/file/d/0B4d0eT ... sp=sharing
by ankan
Sun Aug 14, 2016 6:29 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: yet another attempt on Perft(14)
Replies: 33
Views: 10462

Re: yet another attempt on Perft(14)

Gee :-) Do you have some occupancy profile data? (ALU vs Memory busy) Ho many threads per gpu do you run? How does Pascal perform on 64 bit integers compared to Maxwell/Kepler? -- Srdja From the CPU, I launch a kernel with just a single thread. It uses cuda dynamic parallelism to launch multiple ke...
by ankan
Sun Aug 14, 2016 6:02 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: yet another attempt on Perft(14)
Replies: 33
Views: 10462

Re: The bottom bits

Again: Peter: 110101101011010011110000110011000101100111100000100110000110010101 Ankan: 110101101011010011110000110011000101100111100000100011100110010101 1. Note that the bottom eleven bits are the same. This strongly suggests that the difference of the totals is NOT due to false positive matches ...
by ankan
Sun Aug 14, 2016 5:49 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: yet another attempt on Perft(14)
Replies: 33
Views: 10462

Re: yet another attempt on Perft(14)

sje wrote:Ankan:

Would it be possible to use your revised 128 bit signature program to check Symbolic's output for the first 400 work units (000-399)?

The results I have for these are at https://dl.dropboxusercontent.com/u/316 ... /PerftSums
Ok, will do.