New engine releases & news 2021

[Chessqueen]

I added Chess960/FRC support in Nalwald 1.12.
Only irrelevant performance improvements (less than 10 points, probably), so no need to test this.
Rare crashes and memory consumption issues have been fixed.
I also added a small easteregg, it's not hard to find . It doesn't work very well on Windows though, and also works best with a Unicode compatible console.
Chess960 I only tested with cutechess.
Also, I noticed that windows executables are up to 10% percent slower than Linux executables. I don't know what's the reason for this, and I don't know if this was also an issue for previous releases, but at least it doesn't seem like the Windows Nalwald 1.12 is slower than Windows Nalwald 1.11, so I don't think this will be a performance regression.

Does it actually Blunder a lot, and what is it estimate Elo ratings ?

[j.t.]

Does it actually Blunder a lot, and what is it estimate Elo ratings ?

I don't think it blunders, though I haven't tested against other Chess960 engines. Elo will be around 2700 CCRL (similar to Nalwald 1.11). I don't know about FRC Elo, as, as I said, I haven't tested against other Chess960 engines.

[Gabor Szots]

deleted

[Gabor Szots]

Elo will be around 2700 CCRL

I think it will be much closer to 1700 - in my view he meant Blunder, not Nalwald.

[algerbrex]

I added Chess960/FRC support in Nalwald 1.12.
Only irrelevant performance improvements (less than 10 points, probably), so no need to test this.
Rare crashes and memory consumption issues have been fixed.
I also added a small easteregg, it's not hard to find . It doesn't work very well on Windows though, and also works best with a Unicode compatible console.
Chess960 I only tested with cutechess.
Also, I noticed that windows executables are up to 10% percent slower than Linux executables. I don't know what's the reason for this, and I don't know if this was also an issue for previous releases, but at least it doesn't seem like the Windows Nalwald 1.12 is slower than Windows Nalwald 1.11, so I don't think this will be a performance regression.

Does it actually Blunder a lot, and what is it estimate Elo ratings ?

Hi, if you meant my engine and not Nalwald then no, hopefully not The name is because blunders are something I’m a little to familiar with in my own chess games. And I thought it’d be funny and ironic one day to have a very strong bot named Blunder.

As far as rating, it’s probably around ~1700 if I had to guess.

[connor_mcmonigle]

I've released a new version (v2.3.0) of Seer here: https://github.com/connormcmonigle/seer ... tag/v2.3.0

It should be significantly stronger than the previous release (v2.2.0) and tested at >= 100 elo in self play. As with all previous versions >= v2.0.0 Seer is trained on unique data generated solely from Seer's search+EGTB and does not rely on training code/inference code etc. derived from other engines. Training code and data generation code can be found at https://github.com/connormcmonigle/seer ... e/selfplay.

I've attempted to simplify my naming conventions for binaries. As with the previous release, the SSE3+nopopcnt binary is marked as unofficial and I would prefer if testers refrained from using it for official rating lists/tournaments.

[AlexChess]

Hi,
thank you, I'll use ssse3 nopopcount, but I'm compatible also with popcount sse3 | sse41 binaries and it would be a bit faster.

Regards, Alex

[Rebel]

I've released a new version (v2.3.0) of Seer here: https://github.com/connormcmonigle/seer ... tag/v2.3.0

It should be significantly stronger than the previous release (v2.2.0) and tested at >= 100 elo in self play. As with all previous versions >= v2.0.0 Seer is trained on unique data generated solely from Seer's search+EGTB and does not rely on training code/inference code etc. derived from other engines. Training code and data generation code can be found at https://github.com/connormcmonigle/seer ... e/selfplay.

I've attempted to simplify my naming conventions for binaries. As with the previous release, the SSE3+nopopcnt binary is marked as unofficial and I would prefer if testers refrained from using it for official rating lists/tournaments.

You are going fast

https://prodeo.actieforum.com/t358p25-g ... sults#4061

[connor_mcmonigle]

I've released a new version (v2.3.0) of Seer here: https://github.com/connormcmonigle/seer ... tag/v2.3.0

It should be significantly stronger than the previous release (v2.2.0) and tested at >= 100 elo in self play. As with all previous versions >= v2.0.0 Seer is trained on unique data generated solely from Seer's search+EGTB and does not rely on training code/inference code etc. derived from other engines. Training code and data generation code can be found at https://github.com/connormcmonigle/seer ... e/selfplay.

I've attempted to simplify my naming conventions for binaries. As with the previous release, the SSE3+nopopcnt binary is marked as unofficial and I would prefer if testers refrained from using it for official rating lists/tournaments.

You are going fast

https://prodeo.actieforum.com/t358p25-g ... sults#4061

Thanks for testing!

[connor_mcmonigle]

Hi,
thank you, I'll use ssse3 nopopcount, but I'm compatible also with popcount sse3 | sse41 binaries and it would be a bit faster.

Regards, Alex

I don't use any SSE4.x instructions explicitly so it really shouldn't have any significant impact on performance. You're running Seer by way of x86 emulation on ARM, correct? Likely a native compile would work better anyways...