TalkChess.com

Like I said, I fixed the contains() macro by putting () around it. After that, it was basically just compiler flags to unroll like crazy.

Sorry, will be stopping the silliness now: Perft(1) N = 9 Kn/s: 6870 Perft(2) N = 81 Kn/s: 72321 Perft(3) N = 778 Kn/s: 102910 Perft(4) N = 7128 Kn/s: 128456 Perft(5) N = 69734 Kn/s: 130163 Perft(6) N = 650435 Kn/s: 128473 Perft(7) N = 6385430 Kn/s: 142481 Perft(8) N = 59931436 Kn/s: 287568 Perft(9)...

Oh, and going totally crazy with compiler flags and removing the unused “value” field: Perft(1) N = 9 Kn/s: 0 Perft(2) N = 81 Kn/s: 0 Perft(3) N = 778 Kn/s: 0 Perft(4) N = 7128 Kn/s: 0 Perft(5) N = 69734 Kn/s: 0 Perft(6) N = 650435 Kn/s: 130087 Perft(7) N = 6385430 Kn/s: 172579 Perft(8) N = 59931436...

With !contains() fixed, vector replaced with an array, and slide set to static (so it can be safely inlined): Perft(1) N = 9 Kn/s: 0 Perft(2) N = 81 Kn/s: 0 Perft(3) N = 778 Kn/s: 0 Perft(4) N = 7128 Kn/s: 0 Perft(5) N = 69734 Kn/s: 0 Perft(6) N = 650435 Kn/s: 92919 Perft(7) N = 6385430 Kn/s: 112025...

!contains() doesn't do what you think it does, due to operator precedence. Don't do macros After that, all of your time is spent in std::vector; use a static array instead, so that you can use the stack instead of all this heap allocation.

Question 1.) Is my understanding correct? Seems correct to me. Suppose we have the move e2e4. There are only two inputs which change: King on e1, Pawn on e2 changes from 1 to 0 King on e1, Pawn on e4 changes from 0 to 1 There is apparently some efficiency gain here claimed in the original paper, bu...

Is the accepted answer really correct? So is there a case where the points are not transferred one in one? The Elo system is a statistical system. You don't “win” or “transfer” points; the statistical estimate of your playing strength merely becomes more accurate in the light of new information. (T...

Computers are fast, and can do 8 million positions in a second or less. No need for lifetimes.

Dann Corbit wrote: ↑

Thu Dec 31, 2020 2:55 pm

Did you try LazySMP?
Easy as falling off of a log and seems to work really well.

It's insane that seemingly small tweaks to what was considered the worst-scaling algorithm (shared hash table) somehow produces the best one.

Ferdy wrote: ↑

Wed Dec 30, 2020 7:48 pm

Ajedrecista wrote: ↑

Wed Dec 30, 2020 12:01 pm

What is j in

Code: Select all

F(i) = Prob(i) + SUM[Prob(j)], with j < i

j is all integers from 0 to i-1, inclusive. So F(i) is cumulative probability.

So F(3) = Prob(3) + SUM[Prob(j)] = Prob(3) + (Prob(0) + Prob(1) + Prob(2)) = Prob(3) + F(2).