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



Chess engine: Viridithas 14.0.0

 

Chess engine: Viridithas   Author:Cosmo
Rating Chess Engines Diary CEDR=3616

v.14.0.0:
This release brings significant improvements to Viridithas' neural network architecture, which is now a 16-input-bucket 8-output-bucket 2048x2 -> 16 -> 32 -> 1 network, codenamed hyperstition.

Upgrades, in chronological order:
Remove internal iterative deepening. (#158)
Make UCI implementation more robust. (#159)
Add cuckoo-hashing-based repetition detection. (#163)
Tighter memory layout for the transposition table. (#164)
Update history heuristic on TT cutoffs. (#166)
Add neural network output buckets. (#167)
Use TT to correct static evaluation in QS. (#170)
Make output nicer on legacy Windows consoles. (#172)
Double king bucket count and increase HL to 2048. (#173)
Perform IIR when a TT-entry is low-quality. (#174)
Only SEE-prune moves that land on threatened squares. (#175)
Upgrade network architecture to three-layer network. (#179)
Add Ponder. (#181)
AVX512 support. (#182)
Elo comparison against v13.0.0
STC:

Elo   | 22.27 +- 6.91 (95%)
Conf  | 8.0+0.08s Threads=1 Hash=16MB
Games | N: 3000 W: 837 L: 645 D: 1518
Penta | [22, 277, 719, 451, 31]
https://chess.swehosting.se/test/7796/
LTC:

Elo   | 45.68 +- 10.59 (95%)
Conf  | 60.0+0.60s Threads=1 Hash=128MB
Games | N: 1002 W: 310 L: 179 D: 513
Penta | [1, 62, 246, 189, 3]
https://chess.swehosting.se/test/7798/
(D)FRC:

Elo   | 4.96 +- 7.36 (95%)
Conf  | 8.0+0.08s Threads=1 Hash=16MB
Games | N: 3010 W: 662 L: 619 D: 1729
Penta | [42, 338, 703, 379, 43]
https://chess.swehosting.se/test/7799/
Thanks as always to everyone who helps me out with viri's development, especially everyone on https://chess.swehosting.se/, you're a great bunch.

Binaries are separated by the microarchitecture levels. Most people should use v3!

v1 is supported on all x86-64 CPUs.
v2 has POPCNT, SSSE3, etc.
v3 has AVX2, BMI, LZCNT, etc.
v4 has AVX512, VNNI, etc.



Comments