Search found 37 matches

by grahamj
Fri Nov 15, 2019 9:17 pm
Forum: Computer Chess Club: General Topics
Topic: GPU rumors 2020
Replies: 7
Views: 818

Re: GPU rumors 2020

I meant the host-device latencies also known as kernel launch overhead, can vary between 5ms and 100ms on different systems for null-kernels, this is primary not caused by the PCIe connection, but (speculation) by the little embedded CPU controller in all GPUs that launches the kernels, I admit the...
by grahamj
Thu Aug 29, 2019 4:34 pm
Forum: Computer Chess Club: General Topics
Topic: Lc0 Evaluation Explanation
Replies: 15
Views: 2473

Re: Lc0 Evaluation Explanation

LC0 chooses the move with the most visits. Searching the explanation for 'visits' may help.
by grahamj
Tue Aug 20, 2019 9:15 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: deepdebt
Replies: 8
Views: 1819

Re: deepdebt

Ok the company is betting on reinforcement learning but other research instititures (univ.) could explore the other avenues that could get us there. But as the author says "Every dollar invested in reinforcement learning is a dollar not invested somewhere else..." I watched neural nets being massiv...
by grahamj
Fri Jun 21, 2019 1:29 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Lazy-evaluation of futures for parallel work-efficient Alpha-Beta search
Replies: 53
Views: 14343

Re: Lazy-evaluation of futures for parallel work-efficient Alpha-Beta search

Thanks for the further details. I won't pretend I understand them yet, but your project sounds interesting. You have at least understood that the main problem with using a GPU for chess is how to implement the search tree.
by grahamj
Fri Jun 21, 2019 9:16 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Lazy-evaluation of futures for parallel work-efficient Alpha-Beta search
Replies: 53
Views: 14343

Re: Lazy-evaluation of futures for parallel work-efficient Alpha-Beta search

I am working on a GPU chess engine, but it is not like yours, and does not use alpha-beta search. I will be interested to see how you get on. I described my ideas here. http://indriid.com/2019/2019-01-06-tinsmith.pdf * Tomasulo's Algorithm -- https://en.wikipedia.org/wiki/Tomasulo_algorithm. Tomasul...
by grahamj
Wed Mar 20, 2019 2:58 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Minimax, noisy evaluations, PUCT
Replies: 1
Views: 851

Minimax, noisy evaluations, PUCT

Minimax is known to have a bias which does not diminish with depth. The paper Bias and pathology in minimax search , 2005, A. Sadikov, I. Bratko, I. Kononenko uses KRK endings to investigate this. Here's a simplified situation which I was able to analyse mathematically. Suppose the branching factor ...
by grahamj
Sun Mar 03, 2019 4:04 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Training the trainer: how is it done for Stockfish?
Replies: 22
Views: 5283

Re: Training the trainer: how is it done for Stockfish?

On why AlphaZero (and hence LC0) opted for a Monte Carlo search... There are no strong chess engines which use alpha-beta pruning alone. All the strong engines use other techniques to do a huge amount of pruning. These other techniques are mostly chess-specific, so no good for DeepMind's purposes. A...
by grahamj
Thu Feb 21, 2019 7:41 am
Forum: Computer Chess Club: General Topics
Topic: superhuman engines on common hardware?
Replies: 16
Views: 3556

Re: superhuman engines on common hardware?

mwyoung wrote:
Thu Feb 21, 2019 2:17 am
Maybe I don't understand the question. But it seems to me we are already running superhuman NN engines on common hardware. All I had to do was go to Micro Center.
Maybe you didn't notice the date of the original post.
by grahamj
Tue Feb 19, 2019 8:41 am
Forum: Computer Chess Club: General Topics
Topic: Lc0 ... the GOOD and the BAD
Replies: 54
Views: 10713

Re: Lc0 ... the GOOD and the BAD

I think Monte Carlo search should actually be much stronger in tactical suites than AB engines as it can cover many more positions quicker... How do you imagine that might work? What algorithm do you have in mind? Oh, I would have no clue about how to do that. But I think that there is very big pot...