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: Simbelmyne 1.9.0

-

 


Simbelmyne - UCI engine  Author:Sam Roelants
Rating CEDR=2474 

v.1.9.0:
What's new
This version brings major improvements to the search (getting close to being a somewhat mature search!), and has seen a decent amount of work go into speeding up the underlying move generation.

Probably the bigger and more impactful change has been that I've set up a personal OpenBench instance, powered by 4 dedicated GCP workers. This has massively improved the speed and ergonomics of testing, and means I can finally test smaller improvements that would have been impossible until now.

You can find the OB instance at https://chess.samroelants.com, if you want a recorded history of my many failures, and sporadic successes!

Simbelmyne v1.9 is around 140 Elo stronger than v1.8 in STC self-play, and a small gauntlet places it at around 3114 Elo.

   # PLAYER              : RATING    POINTS  PLAYED    (%)
   1 Avalanche 1.4       : 3214.0     248.5     425   58.5%
   2 Black Marlin 4.0    : 3206.0     244.0     336   72.6%
   3 Nalwald 17          : 3199.0     281.0     425   66.1%
   4 Stash 30            : 3162.0     270.5     424   63.8%
   5 Princhess 0.16      : 3153.0     261.5     424   61.7%
   6 Koivisto 4.0        : 3138.0     194.0     425   45.6%
   7 Patricia 2.0        : 3135.0     249.5     424   58.8%
   8 Simbelmyne 1.9      : 3114.4    1939.0    4156   46.7%
   9 Avalanche 1.3       : 3085.0     160.5     425   37.8%
  10 Patricia 1.0        : 3053.0     146.5     424   34.6%
  11 Polaris 1.8.1       : 3052.0     161.0     424   38.0%

Added features
🔍 Search
PVS SEE Pruning (7.5 +/- 5.7) (#154)
2-ply conthist (11.89 +/- 6.84) (#236)
Clear next ply's killers (10.90 +/- 6.45) (#239)
Use improving in RFP (5.42 +/- 4.09) (#242)
Use improving in FP (5.43 +/- 4.10) (#245)
Add tactical history (19.86 +/- 9.10) (#244)
Singular extensions (9.14 +/- 5.80) (#250)
Double extensions (11.86 +/- 6.68) (#251)
4-ply conthist (5.40 +/- 4.07) (#252)
Use threat-based history (6.70 +/- 4.73) (#253)
⚖️: Evaluation
Use packed eval (22.7 +/- 15.2) (#233)
🐛 Bugfixes
Stop lmr reduction from overflowing (#241)
Clear conthist beteen games (#247)
Misc
Movegen refactor (13.7 +/- 10.7) (#235)
Speed improvements (19.7 +/- 13.7) (#234)
Generate quiets lazily (#227)
Lazily yield TT move before generating captures (22.8 +/- 15.2) (#225)
See the respective PRs for self-play results where relevant

Choosing a binary
This release comes with precompiled binaries for all major platforms. Because
the engine benefits tremendously from more modern CPU instruction sets, there
are binaries compiled for major instruction sets, following the x86-64
microarchitecture levels
as a naming scheme.

Realistically, on modern hardware (< 10 years old), you should be okay to use the
binaries labeled V3. If the engine crashes within the first seconds, try V2,
and so on.

What's Changed
Move "forced move" logic out of negamax by @sroelants in #224
Lazily yield TT Move from move picker by @sroelants in #225
Add helpers for generating tacticals and quiets separately by @sroelants in #226
Generate quiets lazily (17+/-??) by @sroelants in #227
Only store a single board of checkers by @sroelants in #228
Store threats bitboard on Board by @sroelants in #229
Sort killers in FIFO order by @sroelants in #230
is_legal helper by @sroelants in #231
Use packed eval by @sroelants in #233
Random speed improvements by @sroelants in #234
The great movegen rewrite by @sroelants in #235
Add SEE pruning by @sroelants in #154
Openbench preparations by @sroelants in #238
Add 2-ply conthist by @sroelants in #236
Clear the next ply's killers by @sroelants in #239
Stop lmr reduction from overflowing by @sroelants in #241
Search/improving rfp by @sroelants in #242
Simplify quiet history updates by @sroelants in #243
Add tactical history (19.86+/-9.10) by @sroelants in #244
Use improving in FP by @sroelants in #245
Singular extensions by @sroelants in #250
Double extensions by @sroelants in #251
Search/4 ply conthist by @sroelants in #252
Threat-based history by @sroelants in #253


Simbelmyne 1.8.0 - results:
Rustic-Knight 1.0.07/7+77 Games
Camel 1.6.06.5/7+67 Games
Simplex 0.9.8 JA6/6+66 Games
Spaghet Pesto 1.06/6+66 Games
Beast 15 MPV0.5/6-56 Games
Clover 6.20/6-66 Games
Lynx 1.5.04.5/5+45 Games
Princhess 0.17.02/5-15 Games
Odonata 0.9.01.5/5-25 Games
Wasp 7.001/5-35 Games
Obsidian 12.290/5-55 Games
Clarity 7.0.01/4-24 Games
Stockfish dev-202406081/4-24 Games
Steel Fighter 20.5/4-34 Games
Little Beast 150.5/4-34 Games
Lc0 0.31.0-rc30.5/4-34 Games
Titan 1.1.00.5/4-34 Games
SF-PRO 10.06.20240/4-44 Games
CorChess dev-202406060/4-44 Games
PlentyChess 2.0.00/4-44 Games
Marauders 3.60/4-44 Games

Simbelmyne 1.9.0 download


Comments

Post a Comment

Popular posts from this blog

New Strong Engines Test, by Chess Engines Diary, 2024.04.12

-
Image
  💾  552 (!) games from the tournament download   👍 @chessenginesdiary  Country - Poland, City -  Malbork 🕓 Time 3'+3" 💻HP Pavilion i5-1035G1 8GB RAM 🖬 GUI-Banksia All  CEDR  317.321 games download  (01.04.2024 - 3'+3")  HypnoS 030424, Yuli GM Pro 16, Stockfish 16.1 and ShashChess 35 . These four engines scored the same number of points and placed at the top of the table in this strong tournament. Tech table: Engine KN/move NPS dep/mov time/mov mov/game time/game fails Alexandria 6.1.0 3218 576429 30.2 5.6 56.7 316.2 Berserk 13 4779 841025 36.7 5.7 56.0 318.1 Brainlearn 28 2275 366977 31.4 6.2 43.5 269.4 Caissa 1.18 5492 923486 30.2 5.9 50.4 299.7 Clover 6.1.19 4803 854377 31.4 5.6 62.2 349.4 Cool Iris 12.10 2201 356730 27.6 6.2 51.2 315.5 CorChess 20240331 2253 372774 27.4 6.0 52.2 315.8 Crystal 8 3582 571994 20.3 6.3 47.7 298.8 Fire 9.3 5225 876565 22.0 6.0 55.5 331.1 Fisherov chess monk 1.2 3487 619728 36.0 5.6 60.4 340.0 Hyp
Read more

New chess opening book: M11.2 (bin and ctg)

-
Image
  Book - CTG format Customers who wish to use the CTG format book will need a GUI capable of reading this format, ie. Chessbase GUIs.  Please save the book in your book folder. We strongly recommend that you "Optimize" the book in the Book options dialog so the tournament moves are chosen. The book is very wide even in tournament mode unlike most other books. A type of book format mainly used by chess programs from Chessbase , such as Fritz, Chess Base, Komodo, FatFritz, Houdini , etc. Book - BIN format (  Used by chess GUIs Arena, Banksia and Droidfish) Binary books have some important advantages, especially for computer chess engines: space-efficient, fast on accessing and searching, more useful information.  At the most simple form, they are just a conversion of above text-opening books into binary, in the ready form for software and save text-binary converting steps. However, the major number of top books is built and worked with hash values from transpositions. Often ea
Read more

New version chess engine: Lc0 BT4

-
Image
LcZero - CEDR Rating=3683 vBT4: Big transformer 4. New network architecture which builds off of BT3 by adding two types of auxiliary heads, future heads and categorical value heads. The categorical value heads predict a distribution over values of q rather than a WDL outcome distribution, and the future heads predict the moves that will be played over the next two plies. The hope is that these heads will give additional information to the net to improve training speed. We've also fixed half-precision training, so this model will be larger. BT4 training started in mid-October and is expected to take a few months. It has 15 layers with 1024 embedding size, 32 heads per layer, and 1536 emb size, for roughly a doubling in size over BT3.  Individual statistics:  Lc0 0.30.0 Stockfish 16 5/13 -3 13 Games Booot 7.2 6/10 +2 10 Games Critter 1.6a 8/9 +7 9 Games Dragon 3.2 4/8 +0 8 Games Stockfish 20230729 3.5/8 -1 8 Games ShashChess 33.2 3.5/8 -1 8 Games Fisherov chess monk 1.2 3/
Read more

Dragon NNUE by Komodo Chess - it's free!

-
Image
    Dragon by Komodo Chess Larry Kaufman: We have released Dragon at komodochess.com. Dragon uses NNUE (Neural Network Updated Efficiently) technology, originally developed for the game of shogi. Komodo has a great deal of chess knowledge in its evaluation. Training an NNUE network based on this evaluation was both an advantage and a challenge, requiring experimentation with architectures and data generation of billions of positions. The reinforcement learning phase for Dragon is in its infancy, but is already showing great promise. Dragon is a huge strength improvement over 14.1, the last release, about 197 elo in MCTS mode and 189 elo in standard mode, at CCRL blitz time control on one thread based on direct matches. With four threads, the gains were 156 elo in MCTS and 170 in standard mode. The improvement is due to the embedded neural network providing much more accurate evaluation and also in some sense gaining an extra ply or so of search by “seeing” some tactics that a normal ev
Read more