Commercial Release of Ethereal 13.00 (NNUE) for AVX2 Systems

[CMCanavessi]

It really comes down to how fast the ponderhit is sent. How long does it take Ethereal to start the search once getting go? microseconds? Yeah, its a tight window and I'm actually surprised that I could reproduce it. I don't fault Cutechess here (although they do have broken FRC in older versions, and maybe still not updated!) I accept responsibility when needed

See why it's important that some of us use ponder?

[CMCanavessi]

After a while, I also saw a time loss with Banksia (with the old binary). Now I have replaced the binary with your new compiles (version 13.07 is it?) and I started a tournament with Cute Chess again, this time much shorter games at 10sec + 0.5sec and so far, after 32 games, not a single loss on time, so your fix seems to have solved the issue.

Code: Select all

Score of Ethereal 13.00 x64 vs Slow Chess Blitz 2.5 x64: 15 - 7 - 10 [0.625]
...      Ethereal 13.00 x64 playing White: 8 - 3 - 5  [0.656] 16
...      Ethereal 13.00 x64 playing Black: 7 - 4 - 5  [0.594] 16
...      White vs Black: 12 - 10 - 10  [0.531] 32
Elo difference: 88.7 +/- 105.0, LOS: 95.6 %, DrawRatio: 31.3 %
32 of 100 games finished.

[AndrewGrant]

After a while, I also saw a time loss with Banksia (with the old binary). Now I have replaced the binary with your new compiles (version 13.07 is it?) and I started a tournament with Cute Chess again, this time much shorter games at 10sec + 0.5sec and so far, after 32 games, not a single loss on time, so your fix seems to have solved the issue.

Code: Select all

Score of Ethereal 13.00 x64 vs Slow Chess Blitz 2.5 x64: 15 - 7 - 10 [0.625]
...      Ethereal 13.00 x64 playing White: 8 - 3 - 5  [0.656] 16
...      Ethereal 13.00 x64 playing Black: 7 - 4 - 5  [0.594] 16
...      White vs Black: 12 - 10 - 10  [0.531] 32
Elo difference: 88.7 +/- 105.0, LOS: 95.6 %, DrawRatio: 31.3 %
32 of 100 games finished.

Yeah, Version 13.07. Couple elo gainers in there so a small boost. Nothing too major. Glad to see time no losses. I suppose you are one of very few ponder users, as this is the first I am hearing about it. And that issue has been there since ... 11.8X when I added pondering for chesscom's event. Thanks for the help testing.

[CMCanavessi]

Yeah, Version 13.07. Couple elo gainers in there so a small boost. Nothing too major. Glad to see time no losses. I suppose you are one of very few ponder users, as this is the first I am hearing about it. And that issue has been there since ... 11.8X when I added pondering for chesscom's event. Thanks for the help testing.

Wow, then I guess that Arena (being ultra-slow) wasn't affected by it, as I run all my tournaments in it and never experienced any kind of issue. Just now I was using Cute Chess because of FRC, and the issue appeared.

[Chessqueen]

After a while, I also saw a time loss with Banksia (with the old binary). Now I have replaced the binary with your new compiles (version 13.07 is it?) and I started a tournament with Cute Chess again, this time much shorter games at 10sec + 0.5sec and so far, after 32 games, not a single loss on time, so your fix seems to have solved the issue.

Code: Select all

Score of Ethereal 13.00 x64 vs Slow Chess Blitz 2.5 x64: 15 - 7 - 10 [0.625]
...      Ethereal 13.00 x64 playing White: 8 - 3 - 5  [0.656] 16
...      Ethereal 13.00 x64 playing Black: 7 - 4 - 5  [0.594] 16
...      White vs Black: 12 - 10 - 10  [0.531] 32
Elo difference: 88.7 +/- 105.0, LOS: 95.6 %, DrawRatio: 31.3 %
32 of 100 games finished.

Yeah, Version 13.07. Couple elo gainers in there so a small boost. Nothing too major. Glad to see time no losses. I suppose you are one of very few ponder users, as this is the first I am hearing about it. And that issue has been there since ... 11.8X when I added pondering for chesscom's event. Thanks for the help testing.

Ethereal Qualified as 2nd after Igel 3.0.5 and Rubichess performed better than slowchess, very good performance by the trio, I hope they do NOT hit an impenetrable Stargate WALL in the next division of TCEC
https://tcec-chess.com/

[AlexChess]

SCOOP:
Ethereal 13.00 NNUE github works with Fat Fritz 2.0 NET, so we can test it deeply

Hmm, no. The NNUE code is not even compiled in the Github versions. You may have set a network file, but it does not run.
For the final time: No one has networks that can run in Ethereal except me.

OK, indeed they score almost identical. I remove Ethereal 13.00 github NNUE



Regards, AlexChess

[AndrewGrant]

SCOOP:
Ethereal 13.00 NNUE github works with Fat Fritz 2.0 NET, so we can test it deeply

Hmm, no. The NNUE code is not even compiled in the Github versions. You may have set a network file, but it does not run.
For the final time: No one has networks that can run in Ethereal except me.

OK, indeed they score almost identical. I remove Ethereal 13.00 github NNUE



Regards, AlexChess

they are quite literally identical. You can pass Ethereal any file you want with the NNUE and it will just smile and carry on for the non-NNUE version.

[AlexChess]

I guess the questions would be: What GUI are you using? Do you have any logs of the time losses? Do you see the issue when pondering is disabled?

You can also always increase the MoveOverhead option. IF your machine is slow, or has high latency due to other workloads, this may help. Right now its 300ms which is extremely large, even enough for Syzygy on the HDD, but at tournaments I use like 2500ms.

Using CuteChess GUI at the moment, will try later with Arena, but that will be with normal chess, not FRC. As this happened last night and today, I still have no logs (I had them disabled). I'll see in a couple of days if I can get some for you, I'll run only Ethereal matches. Will try with and without ponder. No TBs also, so that's out of the equation.

Okay, I'll be ready when you have more info. I'm sure we can figure something out, as Ethereal is not known for having these sorts of issues.

Please, check also why I can use only 1 CPU or it crashes, while Ethereal 12.75 works perfectly with all CPUs enabled on the same hardware since 6 months, with every GUI.

Regards, AlexChess

[AndrewGrant]

I guess the questions would be: What GUI are you using? Do you have any logs of the time losses? Do you see the issue when pondering is disabled?

You can also always increase the MoveOverhead option. IF your machine is slow, or has high latency due to other workloads, this may help. Right now its 300ms which is extremely large, even enough for Syzygy on the HDD, but at tournaments I use like 2500ms.

Using CuteChess GUI at the moment, will try later with Arena, but that will be with normal chess, not FRC. As this happened last night and today, I still have no logs (I had them disabled). I'll see in a couple of days if I can get some for you, I'll run only Ethereal matches. Will try with and without ponder. No TBs also, so that's out of the equation.

Okay, I'll be ready when you have more info. I'm sure we can figure something out, as Ethereal is not known for having these sorts of issues.

Please, check also why I can use only 1 CPU on it or it crashes, while Ethereal 12.75 works perfectly with all CPU enabled on the same hardware since 6 months.

Regards, AlexChess

I really would not know where to begin looking at emulation. If you can compile on your own, I'm sure you can produce an ARM binary without much effort, and you'll get a newer version than 13.00 anyway. I don't have toolchains setup for that sort of cross compilation, and lack the ability to test in meaningfully. Although there is some code in the makefile for building for android devices, but I don't officially support that really.

[AlexChess]

OK thank you,
until now I have compiled engines only on Ubuntu ARM64 and Silicon M1, but I'll try.

Regards, AlexChess