Seer

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Gerd Isenberg
Posts: 2246
Joined: Wed Mar 08, 2006 7:47 pm
Location: Hattingen, Germany

Re: Seer

Post by Gerd Isenberg » Mon Oct 19, 2020 8:51 am

Guenther wrote:
Sun Oct 18, 2020 9:23 pm
connor_mcmonigle wrote:
Sun Oct 18, 2020 8:48 pm
Gerd Isenberg wrote:
Sun Oct 18, 2020 5:15 pm
...
Again the appeal to engine authors - please create a separate thread if introducing a new engine or important updates. The new engines 2020 thread is intended for testers. It would be interesting to see whether AMD's PEXT beats the native C PEXT implementation, most use magic bitboards as fallback, specially if people demand 32-bit binaries ;-)
...
Sorry to clutter the new engines 2020 thread. I was uncertain as to whether creating a new thread or posting in the existing new engines thread was preferred. In any case, thanks for your interest in my engine.
No need to say sorry! You did not clutter this thread, which I introduced the same way, as the one for 2019.

Gerd is wrong in the way it would be intended for testers only, it was also created, because of easier doing research
in the future, instead of crawling through 1000s of single posts in the future.
(I am also announcing my compilations of programs sometimes, which are not updated with binaries or just source only)
Ok, but it's no fun to scan through 50++ pages with lot of engine sub-trees.
The CCW engine list is practically dead already, if I once should close my chronology too, other people could create
something reliable new, much easier.
May be I will revive the CPW engine releases page.
OTH Gerd is right, when he asks for additional threads from authors of new programs by themselves.
Thanks

connor_mcmonigle
Posts: 272
Joined: Sun Sep 06, 2020 2:40 am
Full name: Connor McMonigle

Re: Seer

Post by connor_mcmonigle » Mon Oct 19, 2020 4:41 pm

I traced the issue with loading the Weights on Windows to Windows interpreting bytes equaling 26 as indicating the end of the file. I've now added binaries to the release page for version 1.0 and appended the version number to the engine's name. Binaries can now be found here: https://github.com/connormcmonigle/seer ... s/tag/v1.0

Additionally, I thought I'd share two nice games my engine played during some recent test matches (annotated by Lichess).

Against latest FabChess Dev:



Against latest Olithink (5.8.7):

Last edited by connor_mcmonigle on Mon Oct 19, 2020 5:06 pm, edited 3 times in total.

connor_mcmonigle
Posts: 272
Joined: Sun Sep 06, 2020 2:40 am
Full name: Connor McMonigle

Re: Seer

Post by connor_mcmonigle » Mon Oct 19, 2020 4:47 pm

BTW, thanks for adding my engine to CPW Gerd! I've been casually following computer chess for many years now and have read many CPW entries. Consequently, it's pretty surreal to see a write up about my own engine on CPW. One minor note, Seer uses the killer move heuristic in move ordering as well.

User avatar
Guenther
Posts: 4010
Joined: Wed Oct 01, 2008 4:33 am
Location: Regensburg, Germany
Full name: Guenther Simon
Contact:

Re: Seer

Post by Guenther » Mon Oct 19, 2020 6:25 pm

connor_mcmonigle wrote:
Mon Oct 19, 2020 4:41 pm
I traced the issue with loading the Weights on Windows to Windows interpreting bytes equaling 26 as indicating the end of the file. I've now added binaries to the release page for version 1.0 and appended the version number to the engine's name. Binaries can now be found here: https://github.com/connormcmonigle/seer ... s/tag/v1.0

...
Thanks for the binaries. I still would like to create a core2 compile (if you have time for a makefile instead of cmake), may be I can squeeze a bit
more out of Seer for my old quadcore. What flags are used for the generic one currently?

Here is the generic compilation output with current 'weights' file from cmd:

Code: Select all

uci
id name Seer 1.0
id author C. McMonigle
option name Clear Hash type button
option name Threads type spin default 1 min 1 max 512
option name Hash type spin default 128 min 1 max 65536
option name Weights type string
uciok
ucinewgame
isready
readyok
setoption name Weights value save.bin
go infinite
info depth 1 seldepth 1 multipv 1 score cp 0 nodes 0 nps 0 tbhits 0 time 2 pv d2d4 g8f6
info depth 2 seldepth 2 multipv 1 score cp 0 nodes 89 nps 12714 tbhits 0 time 6 pv g1f3 g8f6 h2h3
info depth 3 seldepth 3 multipv 1 score cp 72 nodes 235 nps 21363 tbhits 0 time 10 pv g1f3 g8f6 d2d4 e7e5
info depth 5 seldepth 5 multipv 1 score cp 5 nodes 409 nps 27266 tbhits 0 time 14 pv g1f3 g8f6 d2d4 e7e6
info depth 6 seldepth 6 multipv 1 score cp 82 nodes 602 nps 30100 tbhits 0 time 19 pv g1f3 g8f6 d2d4 e7e6 e2e3 f8b4 c2c3
info depth 7 seldepth 7 multipv 1 score cp 69 nodes 1844 nps 36156 tbhits 0 time 50 pv g1f3 c7c5 a2a3 g8f6
info depth 8 seldepth 8 multipv 1 score cp 76 nodes 5256 nps 40744 tbhits 0 time 128 pv d2d4 d7d5 g1f3 c8f5 e2e3 c7c6 f1e2
info depth 9 seldepth 9 multipv 1 score cp 76 nodes 12400 nps 42176 tbhits 0 time 293 pv d2d4 d7d5 e2e3 a7a6 g1f3 e7e6 f1e2 f8b4 b1c3
info depth 10 seldepth 10 multipv 1 score cp 50 nodes 28163 nps 43461 tbhits 0 time 647 pv g1f3 g8f6 d2d4 e7e6 e2e3 c7c5 c2c4 b8c6 d4c5
info depth 11 seldepth 11 multipv 1 score cp 51 nodes 50028 nps 43922 tbhits 0 time 1138 pv e2e4 b8c6 g1f3 d7d5 e4d5 d8d5 d2d4 d5e4 c1e3 c8f5 c2c3
info depth 12 seldepth 12 multipv 1 score cp 70 nodes 61148 nps 44150 tbhits 0 time 1384 pv e2e4 b8c6 g1f3 d7d5 e4d5 d8d5 d2d4 e7e5 b1c3 d5a2 c3a2 c6d4
info depth 13 seldepth 13 multipv 1 score cp 95 nodes 99475 nps 43514 tbhits 0 time 2285 pv e2e4 e7e6 g1f3 c7c5 d2d4 c5d4 f3d4 g8f6 b1c3 b8c6 d4c6
info depth 14 seldepth 14 multipv 1 score cp 87 nodes 163921 nps 43137 tbhits 0 time 3799 pv e2e4 e7e6 d2d4 d7d5 b1c3 d5e4 c3e4 g8f6 e4f6 d8f6 g1f3 b8c6 c2c3 e6e5
info depth 15 seldepth 15 multipv 1 score cp 110 nodes 228357 nps 43167 tbhits 0 time 5289 pv e2e4 e7e6 d2d4 c7c5 g1f3 c5d4 f3d4 b8c6 b1c3 f8b4 a2a3 b4c3 b2c3
info depth 16 seldepth 16 multipv 1 score cp 80 nodes 528274 nps 42482 tbhits 0 time 12434 pv e2e4 e7e5 g1f3 g8f6 b1c3 b8c6 d2d4 f8b4 d4d5 f6e4 d5c6 e4c3 b2c3 b4c3 c1d2 c3a1
info depth 17 seldepth 17 multipv 1 score cp 80 nodes 729836 nps 42331 tbhits 0 time 17240 pv e2e4 e7e5 g1f3 g8f6 b1c3 b8c6 d2d4 f8b4 d4d5 f6e4 d5c6 e4c3 b2c3 b4c3 c1d2 c3a1 d1a1
info depth 18 seldepth 18 multipv 1 score cp 95 nodes 1880119 nps 41419 tbhits 0 time 45391 pv e2e4 c7c5 g1f3 b8c6 f1b5 e7e6 e1g1 g8f6 e4e5 f6d5 f1e1 f8e7 b1c3 e8g8 c3d5 e6d5
BTW am I correct in assuming you are from US?
I would like to have the correct data in the XB/UCI chronology :)
https://docs.google.com/spreadsheets/d/ ... tHs51MQydM
Last edited by Guenther on Mon Oct 19, 2020 6:26 pm, edited 1 time in total.
https://rwbc-chess.de
'chessqueen' 2018-present, aka: 'George' 2013-2016, 'pichy' 2006-2013, 'Jorge Pichard' 2000-2006 (old forum)
Troll barometer:
https://docs.google.com/spreadsheets/d/ ... KSptBx9AUs

User avatar
Sylwy
Posts: 3735
Joined: Fri Apr 21, 2006 2:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: Seer

Post by Sylwy » Mon Oct 19, 2020 6:26 pm

connor_mcmonigle wrote:
Mon Oct 19, 2020 4:41 pm
I traced the issue with loading the Weights on Windows to Windows interpreting bytes equaling 26 as indicating the end of the file. I've now added binaries to the release page for version 1.0 and appended the version number to the engine's name. Binaries can now be found here: https://github.com/connormcmonigle/seer ... s/tag/v1.0
Superb and interesting engine ! THANK YOU ! In tests ! :wink:

Image

Image

THANK YOU also for this thread, for the great CPW .........and many more to Mr.Gerd Isenberg !

connor_mcmonigle
Posts: 272
Joined: Sun Sep 06, 2020 2:40 am
Full name: Connor McMonigle

Re: Seer

Post by connor_mcmonigle » Mon Oct 19, 2020 6:45 pm

Guenther wrote:
Mon Oct 19, 2020 6:25 pm
...
Thanks for the binaries. I still would like to create a core2 compile (if you have time for a makefile instead of cmake), may be I can squeeze a bit
more out of Seer for my old quadcore. What flags are used for the generic one currently?
...
Sure. I can add a core2 compile in a little bit. For the makefile, I'm prioritizing exploring a new NNUE idea involving taking the game phase into consideration right now. I'm getting similar accuracy during training with a model which should be twice as fast.

In any case, the performance you're seeing is .... disappointing :( Slower than Leela on a medium to high end GPU.
If you want to compile using MINGW64 on Windows and can't get CMake configured, Seer should compile with:

(from the build directory)

Code: Select all

g++ ../src/seer.cc -o seer -march=native --std=c++17 -O3 -fopenmp -fconstexpr-ops-limit=1000000000 -I ../include/ -static 
Guenther wrote:
Mon Oct 19, 2020 6:25 pm
...
BTW am I correct in assuming you are from US?
I would like to have the correct data in the XB/UCI chronology :)
https://docs.google.com/spreadsheets/d/ ... tHs51MQydM
Correct! Thank you!
Last edited by connor_mcmonigle on Mon Oct 19, 2020 7:02 pm, edited 1 time in total.

connor_mcmonigle
Posts: 272
Joined: Sun Sep 06, 2020 2:40 am
Full name: Connor McMonigle

Re: Seer

Post by connor_mcmonigle » Mon Oct 19, 2020 7:02 pm

Sylwy wrote:
Mon Oct 19, 2020 6:26 pm
...
Superb and interesting engine ! THANK YOU ! In tests ! :wink:

Image

Image
Thanks for testing! Could you share the processor you're using to test and the clock speed? I'm just trying to get a rough idea of whether my compiles are performing consistently with my expectations. (with an AMD r5 3600 @ 4.00 Ghz, I get ~400knps)

User avatar
Sylwy
Posts: 3735
Joined: Fri Apr 21, 2006 2:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: Seer

Post by Sylwy » Mon Oct 19, 2020 7:14 pm

connor_mcmonigle wrote:
Mon Oct 19, 2020 7:02 pm
Could you share the processor you're using to test and the clock speed? I'm just trying to get a rough idea of whether my compiles are performing consistently with my expectations. (with an AMD r5 3600 @ 4.00 Ghz, I get ~400knps)
Intel i5-7400 (Kaby Lake)

Image

In the first game of my test Seer 1.0 broke Deep Sjeng c't 2010 (a good blitz chess engine having 2925 Elo CCRL Blitz). My impression is that Seer has over 3000 elo points (at TC = 4 '+ 2 "). An impressive game. Unlike anything else seen so far in the NNUE area.

Image


connor_mcmonigle
Posts: 272
Joined: Sun Sep 06, 2020 2:40 am
Full name: Connor McMonigle

Re: Seer

Post by connor_mcmonigle » Mon Oct 19, 2020 7:32 pm

Awesome. The speed you're observing is about in line with what I would expect then. That's a very nice game. Thanks for sharing. In a 300 game 2+1 match against Glaurung 2.2, Seer tested at about +10 elo a couple of weeks ago. It's gained an estimated 40-50 self play elo since that point in time. I don't do very much testing against other engines, but I would guess it's still well below 3000 elo. It can be remarkably insconsistent at times, especially in endgames as a consequence of being so slow.

User avatar
Sylwy
Posts: 3735
Joined: Fri Apr 21, 2006 2:19 pm
Location: IASI - the historical capital of MOLDOVA
Full name: SilvianR

Re: Seer

Post by Sylwy » Mon Oct 19, 2020 7:37 pm

Seer 1.0 znver1 compile on my PC:
Image

Seer 1.0 skylake compile (installed on my PC):
Image

Seer 1.0 znver2 compile on my PC (AMD nostalgia :wink: ):
Image

Post Reply