Since you’re here...

We hope you will consider supporting us today. We need your support to continue to exist, because good entries are more and more work time. Every reader contribution, however big or small, is so valuable. Support "Chess Engines Diary" even a small amount– and it only takes a minute. Thank you.
============================== My email: jotes@go2.pl



Stockfish 15102510 - new development version very strong chess engine UCI

on
Stockfish - chess engines UCI

Previous version chess engine Stockfish 

Author compilation - Lucasart


Information on the compilation:
Use atomics instead of volatile 

Rely on well defined behaviour for message passing, instead of volatile. Three 
versions have been tested, to make sure this wouldn't cause a slowdown on any 
platform. 

v1: Sequentially consistent atomics 

No mesurable regression, despite the extra memory barriers on x86. Even with 15 
threads and extreme time pressure, both acting as a magnifying glass: 

threads=15, tc=2+0.02 
ELO: 2.59 +-3.4 (95%) LOS: 93.3% 
Total: 18132 W: 4113 L: 3978 D: 10041 

threads=7, tc=2+0.02 
ELO: -1.64 +-3.6 (95%) LOS: 18.8% 
Total: 16914 W: 4053 L: 4133 D: 8728 

v2: Acquire/Release semantics 

This version generates no extra barriers for x86 (on the hot path). As expected, 
no regression either, under the same conditions: 

threads=15, tc=2+0.02 
ELO: 2.85 +-3.3 (95%) LOS: 95.4% 
Total: 19661 W: 4640 L: 4479 D: 10542 

threads=7, tc=2+0.02 
ELO: 0.23 +-3.5 (95%) LOS: 55.1% 
Total: 18108 W: 4326 L: 4314 D: 9468 

As suggested by Joona, another test at LTC: 

threads=15, tc=20+0.05 
ELO: 0.64 +-2.6 (95%) LOS: 68.3% 
Total: 20000 W: 3053 L: 3016 D: 13931 

v3: Final version: SeqCst/Relaxed 

threads=15, tc=10+0.1 
ELO: 0.87 +-3.9 (95%) LOS: 67.1% 
Total: 9541 W: 1478 L: 1454 D: 6609 

Resolves #474 


JCER=3256

 Stockfish 15102510 - download

Comments

Post a Comment