TalkChess.com

Regarding: "Very simple. MCTS is the only viable approach given the 80000 nps they got using 4 TPUs." Well, A/B fully replaced any other scheme when computers where much slower than 80Knps, and MCTS never won above A/B in any period of chess engine history...so maybe not so simple :-) Regrading: "it...

As a programmer I am inclined to support elegant solutions, more than patchworks. The basic question is why AlphaZero (and hence LC0) opted for a Monte Carlo search instead of an Alpha/Beta scheme. This is the question that should be answered to have an understanding of the topic, before diving into...

I had forgot I wrote that post on reddit :-) The above explanation is correct, let me just add that the most similar thing to training we do (but it is not training) is automatic tuning, that is used for parameter tweaks, i.e. to find the best value of a set of parameters. This is done in automatic ...

You could use the 'tune' branch: https://github.com/official-stockfish/Stockfish/tree/tune It has some code to allow automatic tuning of a variable in fishtest changing it's value at every game by means of an UCI option. It has no slowdowns for non const parameters, and a possible small slow down fo...

If x <= MAX_MOVES happens to be correct, then make it wrong fixing the definition somewhere else: it should really be wrong

In other words if one is good at what they do stick to it. And if one is not good at it then modify one's approach to get better. :!: In other word use your brain. Usually it is better than the handbook. If, given 2 pieces of code you are able to tell which is the best one, then this is already a g...

I have done it ! :-) Finally I was able to enable performance counters on my VMware virtual Linux and run the amazing perf tool. This is the magical command line: sudo perf stat -r 5 -a -B -e cycles:u ./stockfish bench > /dev/null It _counts_ the CPU instructions of a bench run, repeating the measur...

This being said, I miss hanging out with chess programmers. Hi Tord! nice to see your writing again! Chess programmers miss you too....stockfish devs miss you even more :-) Looking at this discussion it seem to me like looking at a living fossil! People talking about winners, competition, rules.......

What about this one?

http://www.brendangregg.com/perf.html#CPUstatistics

This could be the final weapon! In particular look at the table in the above link:

See also https://perf.wiki.kernel.org/index.php/Tutorial

We have take some baby steps toward this with live displays online of the games. This is not an easy thing to do, I am not sure you have the expertise to do this, you need people that eats burgers and HTML for breakfast :-) If I was the chief of your organization I would have _already_ knocked at, ...