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.
-
peter
- Posts: 2203
- Joined: Sat Feb 16, 2008 6:38 am
- Full name: Peter Martan
Post
by peter » Mon Sep 07, 2020 9:21 pm
Hi Stéphane!
snicolet wrote: ↑Mon Sep 07, 2020 8:51 pm
Hi Volodymyr,
I would like to use the "Night Nurse 0.2" net in fishtest for some tests in Stockfish, do you give me authorization to do so? The idea I have is to run the current Stockfish master net against the current Stockfish with Night Nurse 0.2 (I anticipate that Stockfish with Night Nurse 0.2 will be somewhat weaker, because the search is not optimized for this net), to estimate the Elo difference and maybe work on a contempt implementation for NNUE evals.
Thanks in advance,
Stéphane Nicolet
viewtopic.php?p=856741#p856741
https://github.com/dkappe/NetRunner/
Sorry for answering questions that were not adressed at me, dear Volodymyr regards
Peter.
-
dkappe
- Posts: 859
- Joined: Tue Aug 21, 2018 5:52 pm
- Full name: Dietrich Kappe
Post
by dkappe » Mon Sep 07, 2020 9:24 pm
snicolet wrote: ↑Mon Sep 07, 2020 8:51 pm
Hi Volodymyr,
I would like to use the "Night Nurse 0.2" net in fishtest for some tests in Stockfish, do you give me authorization to do so? The idea I have is to run the current Stockfish master net against the current Stockfish with Night Nurse 0.2 (I anticipate that Stockfish with Night Nurse 0.2 will be somewhat weaker, because the search is not optimized for this net), to estimate the Elo difference and maybe work on a contempt implementation for NNUE evals.
Thanks in advance,
Stéphane Nicolet
Why not use Dark Horse 0.2? It’s maybe within 30 elo of sf12 with proper scaling.
https://www.patreon.com/posts/dark-horse-0-2-40419722
-
carldaman
- Posts: 2081
- Joined: Sat Jun 02, 2012 12:13 am
Post
by carldaman » Mon Sep 07, 2020 11:04 pm
dkappe wrote: ↑Mon Sep 07, 2020 9:24 pm
snicolet wrote: ↑Mon Sep 07, 2020 8:51 pm
Hi Volodymyr,
I would like to use the "Night Nurse 0.2" net in fishtest for some tests in Stockfish, do you give me authorization to do so? The idea I have is to run the current Stockfish master net against the current Stockfish with Night Nurse 0.2 (I anticipate that Stockfish with Night Nurse 0.2 will be somewhat weaker, because the search is not optimized for this net), to estimate the Elo difference and maybe work on a contempt implementation for NNUE evals.
Thanks in advance,
Stéphane Nicolet
Why not use Dark Horse 0.2? It’s maybe within 30 elo of sf12 with proper scaling.
https://www.patreon.com/posts/dark-horse-0-2-40419722
I salute Stéphane's idea to work on a contempt feature for SF-NNUE, but shouldn't he be asking you, Dietrich [creator of NiNu] for permission?

-
dkappe
- Posts: 859
- Joined: Tue Aug 21, 2018 5:52 pm
- Full name: Dietrich Kappe
Post
by dkappe » Mon Sep 07, 2020 11:07 pm
carldaman wrote: ↑Mon Sep 07, 2020 11:04 pm
dkappe wrote: ↑Mon Sep 07, 2020 9:24 pm
snicolet wrote: ↑Mon Sep 07, 2020 8:51 pm
Hi Volodymyr,
I would like to use the "Night Nurse 0.2" net in fishtest for some tests in Stockfish, do you give me authorization to do so? The idea I have is to run the current Stockfish master net against the current Stockfish with Night Nurse 0.2 (I anticipate that Stockfish with Night Nurse 0.2 will be somewhat weaker, because the search is not optimized for this net), to estimate the Elo difference and maybe work on a contempt implementation for NNUE evals.
Thanks in advance,
Stéphane Nicolet
Why not use Dark Horse 0.2? It’s maybe within 30 elo of sf12 with proper scaling.
https://www.patreon.com/posts/dark-horse-0-2-40419722
I salute Stéphane's idea to work on a contempt feature for SF-NNUE, but shouldn't he be asking you, Dietrich [creator of NiNu] for permission?
It’s publicly available. Why shouldn’t he use it?
-
snicolet
- Posts: 4
- Joined: Thu Jul 05, 2018 8:04 pm
- Full name: Stephane NICOLET
Post
by snicolet » Tue Sep 08, 2020 10:47 am
>> I salute Stéphane's idea to work on a contempt feature for SF-NNUE, but shouldn't he be asking you, Dietrich [creator of NiNu] for permission?

> It’s publicly available. Why shouldn’t he use it?
In fact I did also ask dkappe via Github
https://github.com/dkappe/NetRunner/com ... 403ba7c54d
So I interpret the question "It’s publicly available. Why shouldn’t he use it?" as an authorization, thanks!
Stéphane
-
carldaman
- Posts: 2081
- Joined: Sat Jun 02, 2012 12:13 am
Post
by carldaman » Thu Sep 10, 2020 8:19 am
FYI, Dark Horse 0.2a appears to be the most aggressive nnue net in terms of playing style I've tested so far.
-
voffka
- Posts: 186
- Joined: Sat Jun 30, 2018 8:58 pm
- Location: Ukraine
- Full name: Volodymyr Shcherbyna
-
Contact:
Post
by voffka » Wed Sep 23, 2020 9:15 pm
Igel 2.8.0 at
https://github.com/vshcherbyna/igel/releases/tag/2.8.0 (binaries for Windows).
This release brings major improvements in strength due to a fix of a null move pruning bug. A tiny code change brings +54 elo at LTC (
http://chess.grantnet.us/test/7463/). An updated Night Nurse net from Dietrich Kappe brings additional 10-20 elo.
What's new:
- Fix null move pruning bug (54 elo)
- Use 250cp as lazy eval threshold
- Improve time management for sudden death time controls
- Use Dietrich Kappe's Night Nurse improved net (internal version ninu-0.2.5b.nnue)
Regression tests against Igel 2.7.0 indicate:
+75 elo at 10+0.1
+120 elo at 10+0.0 (sudden death time control)
-
Dokterchen
- Posts: 117
- Joined: Wed Aug 15, 2007 10:18 am
- Location: Munich
Post
by Dokterchen » Fri Sep 25, 2020 7:08 am
Nice improvement! A first quick test shows that Igel is now among top 10 programs.
Torsten
-
Rebel
- Posts: 5811
- Joined: Thu Aug 18, 2011 10:04 am
Post
by Rebel » Fri Sep 25, 2020 7:52 am
Can you provide a pure popcount version without AVX?
90% of coding is debugging, the other 10% is writing bugs.
-
peter
- Posts: 2203
- Joined: Sat Feb 16, 2008 6:38 am
- Full name: Peter Martan
Post
by peter » Fri Sep 25, 2020 8:38 am
Rebel wrote: ↑Fri Sep 25, 2020 7:52 am
Can you provide a pure popcount version without AVX?
Preferably a SSE4.1-popcnt one, well for Xeon X5670?

regards
Peter.