Skip to content

Conversation

@TheRealGioviok
Copy link
Contributor

@TheRealGioviok TheRealGioviok commented Jan 3, 2026

Speedup barely measurable at vstc with low hash. Becomes more evident with more hash / threads. Huge thanks to @KierenP for his large pages impl, which has been simplified (this doesn't implement 1GB huge pages) into what this pr adds.

Passed VSTC with high hash

Test  | 2mbpages
Elo   | 4.47 +- 2.82 (95%)
SPRT  | 5.0+0.05s Threads=1 Hash=128MB
LLR   | 2.95 (-2.94, 2.94) [0.00, 3.00]
Games | N: 18356 W: 4705 L: 4469 D: 9182
Penta | [188, 1985, 4628, 2157, 220]

https://clockworkopenbench.pythonanywhere.com/test/1065/

VSTC with low hash sss

Test  | 2mbpages
Elo   | 0.75 +- 0.89 (95%)
SPRT  | 5.0+0.05s Threads=1 Hash=16MB
LLR   | -0.70 (-2.94, 2.94) [0.00, 3.00]
Games | N: 184172 W: 45872 L: 45473 D: 92827
Penta | [1953, 20626, 46656, 20771, 2080]

https://clockworkopenbench.pythonanywhere.com/test/1064/

Bench: 13398527

@87flowers 87flowers merged commit 4250327 into official-clockwork:main Jan 4, 2026
28 checks passed
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.

2 participants