bigger PAWN_HISTORY_SIZE #6350
bigger PAWN_HISTORY_SIZE #6350
Conversation
bench 2128316
|
interesting, I'm getting a pretty dramatic decrease in nodes per second (~4%) with this branch. Maybe someone else can corroborate? If so that means the increased size of the pawn history table is worth a lot, but also that it might be good to somehow try to optimize it for cache locality or density. |
maybe i can test a decrease in NPS? if so, say pls, how to do it |
Use the speedtest function and compare the before and after. |
the results may be noisy, because while the these tests run, my processor was running a lichess bot :)
so idk are these results true of not
|
|
I think you tested both times the patch. Both times it shows the commit And yes, to ensure that the test is as accurate as possible, nothing CPU-intensive should be running on the system at the same time. |
ok, i'll rerun tests, but if you think it's the same version, why the results are different? |
yeah, i think it's true, exactly 4%
|
|
master patch no slowdown here |
|
what is label "gainer"? |
the best label :-) |
|
I think I might have just measured the wrong binary... looks good to me now :P |
|
Master: Patch: i tried second time |
a pending SF patch requires an update of the memory estimate official-stockfish/Stockfish#6350
As the worker data is quite large (28MB after #6350) we can make use of huge pages as a speedup. prior to #6350 STC passed elo gaining bounds: LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 166272 W: 43479 L: 42993 D: 79800 Ptnml(0-2): 540, 17598, 46365, 18102, 531 https://tests.stockfishchess.org/tests/view/68e9f3c0d323fd15c04e3ba4 Tested the speedup on a large machine with speedtest: ==== master ==== Average (over 20): 288644510 ==== largePageWorker ==== Average (over 20): 292082422 Test after #6350: ==== rustam-cpp-testPR ==== Average (over 20): 291035351 ==== rustam-cpp-testPR-pages ==== Average (over 20): 291937367 #6359 No functional change
As the worker data is quite large (28MB after official-stockfish#6350) we can make use of huge pages as a speedup. prior to official-stockfish#6350 STC passed elo gaining bounds: LLR: 2.95 (-2.94,2.94) <0.00,2.00> Total: 166272 W: 43479 L: 42993 D: 79800 Ptnml(0-2): 540, 17598, 46365, 18102, 531 https://tests.stockfishchess.org/tests/view/68e9f3c0d323fd15c04e3ba4 Tested the speedup on a large machine with speedtest: ==== master ==== Average (over 20): 288644510 ==== largePageWorker ==== Average (over 20): 292082422 Test after official-stockfish#6350: ==== rustam-cpp-testPR ==== Average (over 20): 291035351 ==== rustam-cpp-testPR-pages ==== Average (over 20): 291937367 official-stockfish#6359 No functional change
5 participants
STC (10+0.1 th1) was accepted:
LLR: 2.95 (-2.94,2.94) <0.00,2.00>
Total: 75712 W: 19701 L: 19326 D: 36685
Ptnml(0-2): 254, 8738, 19513, 9081, 270
https://tests.stockfishchess.org/tests/view/68e286d5fa806e2e8393d160
LTC (60+0.6 th1) was accepted:
LLR: 2.96 (-2.94,2.94) <0.50,2.50>
Total: 108492 W: 28068 L: 27604 D: 52820
Ptnml(0-2): 60, 11639, 30390, 12091, 66
https://tests.stockfishchess.org/tests/view/68e3e564a017f472e763dac0
bench 2128316