Each patch was within the error bar. There is currently a bug with razoring PV nodes, which needs fixing. I will reschedule a test with fixed razoring.syzygy wrote:In the end the result wasn't that great, it seems?lucasart wrote:Completing the picture with newer tests:
Razoring at PV nodes
http://tests.stockfishchess.org/tests/v ... 5c4de1c0b9
http://tests.stockfishchess.org/tests/v ... 763e075658
Futility + LMP + SEE pruning at PV nodes in search()
http://tests.stockfishchess.org/tests/v ... 28c6bc66fd
http://tests.stockfishchess.org/tests/v ... 26a5666bc6
Futility pruning + SEE pruning at PV nodes in qsearch()
http://tests.stockfishchess.org/tests/v ... 17fe3fc6db
http://tests.stockfishchess.org/tests/v ... 28c6bc6704
Probcut at PV nodes
http://tests.stockfishchess.org/tests/v ... 6d8243c853
http://tests.stockfishchess.org/tests/v ... 5c4de1c0be
Null move pruning at PV nodes
http://tests.stockfishchess.org/tests/v ... 64db6b6d6f
http://tests.stockfishchess.org/tests/v ... 763e07564f
Child node Futility pruning at Root (aka. reverse futility pruning, beta pruning, static null move pruning)
http://tests.stockfishchess.org/tests/v ... 64db6b6d6d
http://tests.stockfishchess.org/tests/v ... 7a947a50bb
http://tests.stockfishchess.org/tests/v ... 047028beb6
Even with the odds considerably favouring the "simplification", it seems to have failed.
Also, the test you're looking at does not have Futility/SEE/LMP. So far, no test has been scheduled with the whole lot at once.