Repository navigation
Estimated Elo
3000 CCRL 40/15
ELO | 256.55 +- 9.83 (95%)
CONF | 8.0+0.08s Threads=1 Hash=32MB
GAMES | N: 5024 W: 3502 L: 346 D: 1176
https://chess.swehosting.se/test/3641/
ELO | 232.69 +- 13.03 (95%)
CONF | 40.0+0.40s Threads=1 Hash=128MB
GAMES | N: 2500 W: 1634 L: 172 D: 694
https://chess.swehosting.se/test/3640/
Author's Note
During my time spent developing version 4 I became very interested in king safety evaluation features, and as a result dramatically shifted the way that Wahoo plays. Based on my observations, it does a much better job of keeping its own king safe, while also exploiting enemy weaknesses much more cleanly. I also developed an experimental version that replaced HCE king safety with a small neural network. I may release this version separately at some point in time, as it did result in an improvement of some 24ish Elo.
New Engine Features
- Use More Increment: +6 Elo #118
- Use TT Score as Static Eval: +14 Elo #119
- RFP before NMP: +11 Elo #120
- Nonlinear King Safety Re-write: +41 Elo #123
- King Rank (Safety): +12 Elo #125
- Xray Mobility and Safety: +14 Elo #126
- More Safety Params: +10 Elo #128
- Piece Tropism (Safety): +15 Elo #130
- Pawn Storm (Safety): +6 Elo #132
- SEE Pruning: +54 Elo #135
- SEE Move Ordering: +16 Elo #136
- Exclude Qsearch Losing Captures: +17 Elo #137
- Late Move Pruning: +27 Elo #138
- Rule of Square for Passers: +11 Elo #140
- Prevent Unsound Pruning when Mate is Found: +7 Elo #142
- New LMP Formula: +35 Elo #143
- Improving in LMP: +21 Elo #146
- Improving in RFP: +10 Elo #147
- Safety-Related Pawn Structure: +9 Elo #159
- Safety Pawn Structure Improvement: +5 Elo #160
- Queen Contact Checks (Safety): +10 Elo #161
- Side-to-move Relative Threats: +4 Elo #162
- Mirror PSTs: +9 Elo #164
- Node Time Management: +10 Elo STC #167
- Opposite Colored Bishop Endgame Handling: #168
Bugfixes
- Wahoo no longer reports bogus mate scores due to broken mate distance pruning
Binaries
Wahoo's release scheme follows the microarchitecture levels.
x86_64-v1 is the slowest but compatible with almost anything.
x86_64-v2 is faster but is compatible with CPUs pre-Haswell/Excavator.
x86_64-v3 is faster still and recommended on modern systems.
x86_64-v4 is the fastest but requires AVX-512 support.