New Engine: Weiawaga

[Heiaha]

Hi all, hope your holidays have been nice and you're keeping warm if you're in the North.

I'd like to submit my engine today for inclusion in CCRL. It's written in Rust, uses the UCI protocol, is open source, and its repository can be found here:

https://github.com/Heiaha/Weiawaga

I just published a new release a few hours ago where you can find the binaries. Alternatively if you have cargo installed you can very easily build yourself by running "cargo build --release" in its top directory. My estimation right now is that on a single thread it would be around 2500 Elo on CCRL and perhaps ~180-200 more using 4 threads.

Thanks very much in advance, and please let me know if there's any information I can provide to help.

[bastiball]

Congrats for the new engine and for the release!

[Heiaha]

Thanks very much!

[Gabor Szots]

Seems a nice new engine. I would have welcomed some words on its web page about its past and future, apart from having been a java engine originally.
I could build it myself but your exe is somewhat faster.

Congratulations!

[Gabor Szots]

Seems a nice new engine. I would have welcomed some words on its web page about its past and future, apart from having been a java engine originally.
I could build it myself but your exe is somewhat faster.

Congratulations!

BTW, your name and country? Needed for the list.

[Heiaha]

Hi Gabor,

Yes I've been planning on expanding the README, so perhaps I'll include some of that info. Also, my name is Kyle Krowpman and I'm in the US.

[Gabor Szots]

Thanks Kyle. I do the asking, Basti does the testing.

[bastiball]

Just finished the testing, Gabor might add some more testing

[Gabor Szots]

Just finished the testing, Gabor might add some more testing

Not sure. I am not a Wagner fan.

[Heiaha]

I pinky swear no Valkyries will emerge from your computer if you decide to.

[Gabor Szots]

I pinky swear no Valkyries will emerge from your computer if you decide to.

I have no objection to Valkyries if Wotan is not present.

Oh, and I forgot: excellent choice of name!

[Heiaha]

Haha thank you, I'm glad you like it.

[Gabor Szots]

Just running it under Arena and I've noticed it does not display any PV, only the best move when it moves. Not that it's the only engine behaving so but most of them send their PV output at all depths (or at least starting from a certain depth).

[Heiaha]

Hi Gabor,

That's strange, I just tested the binary I have on Github for release 4 and it displays the pv for me in terminal, CuteChess, and Arena. I have it using the transposition table for getting the pv which I know isn't ideal, but it should definitely give you more information than just the best move once the search finishes. Here's an example from the starting position:

Weiawaga v4.0
https://github.com/Heiaha/Weiawaga
uci
id name Weiawaga
id author Heiaha
option name Hash type spin default 16 min 1 max 131072
option name Threads type spin default 1 min 1 max 512
uciok
go depth 10
info currmove b1c3 depth 1 seldepth 1 time 0 score cp 22 nodes 21 nps 21000 pv b1c3
info currmove b1c3 depth 2 seldepth 2 time 0 score cp 17 nodes 66 nps 66000 pv b1c3 b8c6
info currmove b1c3 depth 3 seldepth 4 time 0 score cp 22 nodes 374 nps 374000 pv b1c3 g8f6 g1f3
info currmove b1c3 depth 4 seldepth 6 time 1 score cp 17 nodes 668 nps 334000 pv b1c3 g8f6 g1f3 b8c6
info currmove b1c3 depth 5 seldepth 7 time 1 score cp 13 nodes 1504 nps 752000 pv b1c3 g8f6 g1f3 b8c6 d2d4
info currmove b1c3 depth 6 seldepth 10 time 2 score cp 17 nodes 2854 nps 951333 pv b1c3 g8f6 g1f3 b8c6 d2d4 d7d5
info currmove b1c3 depth 7 seldepth 11 time 3 score cp 12 nodes 6360 nps 1590000 pv b1c3 g8f6 g1f3 b8c6 d2d4 d7d5 c1e3
info currmove g1f3 depth 8 seldepth 15 time 6 score cp 17 nodes 13514 nps 1930571 pv g1f3 b8c6
info currmove e2e4 depth 9 seldepth 15 time 24 score cp 17 nodes 84165 nps 3366600 pv e2e4 c7c6
info currmove e2e4 depth 10 seldepth 19 time 40 score cp 23 nodes 75222 nps 1834682 pv e2e4 b8c6 g1f3 g8f6
info score cp 23
bestmove e2e4

[Gabor Szots]

I have found the reason: UCI filter was in use under Arena. When I switch it off the PV's are displayed correctly.
Only, I always use UCI filter because it is useful for most engines and cannot be set individually for each engine.

UCI Filter" lets Arena ignore useless information that is sent by engines during the first 5 seconds of a search. Important information like mainlines and the best moves will of course always be processed. During the first 5 seconds many engines send a lot of information to Arena, which may become very slow because of this. This has a big impact especially at short time controls.

So it seems Arena does not think the PV is important information.

[Ray]

Hi all, hope your holidays have been nice and you're keeping warm if you're in the North.

I'd like to submit my engine today for inclusion in CCRL. It's written in Rust, uses the UCI protocol, is open source, and its repository can be found here:

https://github.com/Heiaha/Weiawaga

Graham says that this plays FRC, but it does not have a UCI_Chess960 UCI parameter so I'm not sure that it does. Could you confirm ?

[Graham Banks]

91ST AMATEUR SERIES (Division 8 )

Intel i7 Quad
ChessGUI
256mb hash each
3-4-5 piece tablebases
Ponder off
GM2700+.cgb book
40 moves in 16 minutes repeating (adapted for the CCRL)
4 cycles 44 rounds
All engines 64-bit where available

The top engine will gain automatic promotion. Any other changes of division will be at my discretion and may not necessarily involve playoff matches.


Participants

Weiawaga 4.0 64-bit (new)
Loki 3.5.0 64-bit (relegated)
Waxman 2020
MinimalChess 0.6 64-bit
Cosette 5.1 64-bit
NoC 4.20 64-bit
Popochin 4.2 64-bit
Ares GB 1.1 64-bit
Tantabus 1.0.2 64-bit (new)
Isa 2.0.83 64-bit
Barbarossa 0.6.0 64-bit
Anka 0.6.3 64-bit


Games and standings will be available after every six rounds from the following link:
viewtopic.php?f=7&t=13836

Web based link for live viewing (huge thanks to Jay - Berserk author).
http://tlcv.net/16092

Alternatively, if you install TLCV (Tom's Live Chess Viewer) on your computer, you can watch the games live move by move. You'll also be able to chat to others following the tournament in the chatroom there.
download/file.php?id=42959
Host - GrahamCCRL.dyndns.org Port - 16092

Linux users can use Livius:
https://github.com/kmar/livius

There is also a Livius windows version.
It has live pv boards as a nice addition.
http://www.crabaware.com/livius/

[Heiaha]

Hi,

No it doesn't have FRC support. Maybe that's something I'll add in the future, but for now if you tried it would generate extremely wrong moves related to castling.

[Heiaha]

Hi all,

I've released a new version of Weiawaga, and you can find some of the binaries here:

https://github.com/Heiaha/Weiawaga/releases/tag/v5.0

Mostly this release introduces a network-based evaluation based on training via self-play games with the previous version of Weiawaga. The search is largely structurally the same, with the exception of the introduction of reverse futility pruning. I estimate that it is 350 +/- 17.7 Elo stronger than the last version of Weiawaga. I'd greatly appreciate any further testing the CCRL testers would do .

[Graham Banks]

Hi all,

I've released a new version of Weiawaga, and you can find some of the binaries here:

https://github.com/Heiaha/Weiawaga/releases/tag/v5.0

Mostly this release introduces a network-based evaluation based on training via self-play games with the previous version of Weiawaga. The search is largely structurally the same, with the exception of the introduction of reverse futility pruning. I estimate that it is 350 +/- 17.7 Elo stronger than the last version of Weiawaga. I'd greatly appreciate any further testing the CCRL testers would do .

Very good.
I'll start a gauntlet on Monday.

[Heiaha]

Thanks very much! Have a great weekend.

[Graham Banks]

Playing in Division 5 now.
https://tlcv.net/16091

[Graham Banks]

Gauntlet live in 16069.