Search found 3283 matches

by Daniel Shawul
Sat Jul 20, 2019 1:16 pm
Forum: Computer Chess Club: General Topics
Topic: buying a new computer
Replies: 165
Views: 4923

Re: buying a new computer

Success at last!! … … ... I did a quick benchmark on the CPU (ryzen 9 3900x) and GPU (RTX 2070 super). Stockfish seems to scale linearly across the 12 cores with its lazy smp implementation. So i got about 1.8 mnps on 1 core using latest source compiled with gcc 7.4, and get 21 mnps using all 12 co...
by Daniel Shawul
Fri Jul 19, 2019 3:24 pm
Forum: Computer Chess Club: General Topics
Topic: How to get ScorpioNN to work?
Replies: 6
Views: 512

Re: How to get ScorpioNN to work?

Thanks Torsten! I didn't see your reply until now. So your GPU is an RTX 2070 and you already got 20k and 28k with FP16 and INT8 with the maddex net. I just bought an RTX 2070-super and I am not able to get good numbers with FP16 (just 10k) for some reason with maddex net but INT8 gives about 33k. W...
by Daniel Shawul
Fri Jul 19, 2019 2:54 pm
Forum: Computer Chess Club: General Topics
Topic: buying a new computer
Replies: 165
Views: 4923

Re: buying a new computer

Well it turns out lc0 performs well on the rtx 2070-super cards getting nps of FP16 28k nps FP32 10k nps So this must have been a problem with TensorRT library i am using in scorpio. I used to get comparable nps with FP16 and FP32 as lc0 on a volta chip so I am not sure what is going on here. Howev...
by Daniel Shawul
Fri Jul 19, 2019 1:48 pm
Forum: Computer Chess Club: General Topics
Topic: buying a new computer
Replies: 165
Views: 4923

Re: buying a new computer

I don't think lc0 consumes that much memory and from my calculations it could go upto 17 hours analysis with just 12 gb ram. Thats one of the reason i decided to stick with 16 gb ram. Daniel 1 node takes 250 bytes, or in other words 1GB is needed for 4M nodes. (and also one NN cache entry takes 350...
by Daniel Shawul
Fri Jul 19, 2019 1:26 pm
Forum: Computer Chess Club: General Topics
Topic: buying a new computer
Replies: 165
Views: 4923

Re: buying a new computer

I don't think lc0 consumes that much memory and from my calculations it could go upto 17 hours analysis with just 12 gb ram. Thats one of the reason i decided to stick with 16 gb ram. Daniel 1 node takes 250 bytes, or in other words 1GB is needed for 4M nodes. (and also one NN cache entry takes 350...
by Daniel Shawul
Fri Jul 19, 2019 2:49 am
Forum: Computer Chess Club: General Topics
Topic: buying a new computer
Replies: 165
Views: 4923

Re: buying a new computer

I am looking for a new computer since it has been a while. I saw a 2990wx with 16 GB ram, 2080ti, 500 SSD drive for 3k. Overall do you think this is a reasonable deal? Will this be a good setup for a few years? I know new chips are coming out, but will this be good "bang for your buck"? I know a fe...
by Daniel Shawul
Fri Jul 19, 2019 2:12 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Self-taught AI solves Rubik's cube
Replies: 15
Views: 501

Re: Self-taught AI solves Rubik's cube

Thanks Peter. I did not buy the paper, but from reading the first page: - They use weighted A* search. - They use something called "deep approximate value iteration" to train the heuristic function used by the A* algorithm. In regular value iteration you use a big lookup table and compute the table ...
by Daniel Shawul
Fri Jul 19, 2019 2:00 am
Forum: Computer Chess Club: General Topics
Topic: buying a new computer
Replies: 165
Views: 4923

Re: buying a new computer

Well it turns out lc0 performs well on the rtx 2070-super cards getting nps of FP16 28k nps FP32 10k nps So this must have been a problem with TensorRT library i am using in scorpio. I used to get comparable nps with FP16 and FP32 as lc0 on a volta chip so I am not sure what is going on here. Howeve...
by Daniel Shawul
Thu Jul 18, 2019 3:32 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Self-taught AI solves Rubik's cube
Replies: 15
Views: 501

Re: Self-taught AI solves Rubik's cube

Dumb reporter from jakarta post In a world first, researchers at the University of California have developed a computer algorithm that can solve a Rubik's Cube without a neural network, machine learning techniques, "specific domain knowledge," or human assistance. And this By successfully being able...