Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Do lmr for more captures. #2420

Closed

Conversation

Vizvezdenec
Copy link
Contributor

passed STC
http://tests.stockfishchess.org/tests/view/5dd4d51df531e81cf278eaac
LLR: 2.97 (-2.94,2.94) [-1.50,4.50]
Total: 16495 W: 3591 L: 3434 D: 9470
passed LTC
http://tests.stockfishchess.org/tests/view/5dd52265f531e81cf278eace
LLR: 2.96 (-2.94,2.94) [0.00,3.50]
Total: 23598 W: 3956 L: 3716 D: 15926
based on machinery introduced by vondele. Logic behind patch if relatively simple - if we reduce less with high hitrate of TT somewhat logical is to reduce more with low hitrate. For example enable all captures for LMR.
Constant is arbitrary and can be tweaked. :)
bench 5183523

bench 5183523
@Alayan-stk-2
Copy link

Very good performance for this patch, but I wonder about something with this patch and vondele's : how does it behave in deep searches ?

If someone uses a bigger than optimal TT size, it's quite easy to reduce it to match the TC. But for long searches, one often cannot use a bigger hash. Do we know how the tthitrate would behave in this situation ?

@ddugovic
Copy link

If we do end up changing this, next time could it be tried without a divide operation?
8 * thisThread->ttHitAverage < 3 * ttHitAverageResolution * ttHitAverageWindow

@anshulongithub
Copy link

Many congratulations!! to one and only @my :)

@vondele
Copy link
Member

vondele commented Nov 21, 2019

@ddugovic the divide can actually be optimized at compile time (i.e. the whole right hand side is a constant) and so is for free.

@snicolet snicolet closed this in 3f41913 Nov 21, 2019
@snicolet
Copy link
Member

Merged via 3f41913, congrats!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants