TalkChess.com

Yeah, even SF8 doesn't have too hard of a time finding the win after Rf8 is played if you just start the engine on that position (finds it at iteration 38 just like in your posted example). FEN: 5rkq/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 w - - 0 1 Stockfish_8_x64: 25/47 00:02 4,354k 1,882k 0.00 Qh4-f4 d...

Yeah, if SF8 is just dumped in this position, on my machine (using just 1 core) it consistently sticks with Rf8 at 0.00 until d35 (takes about 30 seconds on my machine) across a bunch of runs. However, if I let it play from a couple moves back, so the hash is populated when it arrives at this positi...

Because the games for Table 2 were not just organically played games. They trained the neural nets with self play, and once training was done they took the 12 most popular openings according to their measure in human play and had SF and AlphaZero play matches from the resulting positions of each 12 ...

Same problem, though. Hard for SF to get into the 2...e6 Sicilian as black against an opponent that opens almost exclusively 1.d4 :) Also, SF's score as black in that opening was still not particularly good; most of the reason AlphaZero's overall score in the B40 games was so low was because of the ...

Well, it doesn't follow that an anti book would be a piece of cake. The example opening you picked where it performed relatively (heavy emphasis on "relatively") poorly is one you couldn't force no matter what book you gave SF. No amount of book magic will let SF force an opponent that always meets ...

Just something to be aware of, there's probably some selection bias here, since SF only outputs hashfull after 1 second of searching. This means that you'll only see hashfull on lower depths when it's perhaps failed high/low or otherwise had to spend a bunch of time for other reasons; it may well ha...

Yes

Rybka 4.1 is the latest version.

Any comments on the TCEC 10 time controls? Are they using a 20 core server? How much RAM? It's all in the popup from the "Rules" button on the site (all the following just copy/paste): CPUs: 44 Cores -> 2 x Intel Xeon E5 2699 v4 @ 2.8 GHz Motherboard: Supermicro X10DRL-i RAM: 64 GB DDR4 ECC SSD: Cr...

Indeed. That is a better type of test in general, and I had done some of those as well. I posted what I did to make it as easy to reproduce as possible, not that your example is all that hard to work with :) At any rate, for completeness I did quick runs on this position, doing what I had to get it ...

Interesting timing! I was just getting ready to post something about this as well. I had noticed the same thing, but wasn't sure if I was doing something wrong (although I found that unlikely, since the exact same steps worked with smaller hash sizes). With hash 1024 MB, single thread, initial posit...