Senpai 1.0 (new engine)

[j_romang]

Thanks Lucas ; but I would also be happy to follow Fabien's work and progress until Senpai 2.0 Fabien, do you think you will publish your work in a public version control system ?

[JuLieN]

Thanks Lucas ; but I would also be happy to follow Fabien's work and progress until Senpai 2.0 Fabien, do you think you will publish your work in a public version control system ?

A big part of the fun of developing a chess engine is to prepare a big "coup" by releasing an engine hundreds of Elo points stronger than its previous version... The surprise element is very fun... And this is something that such a public development system prevents. Therefore, not all engines author want to go the Stockfish way.

[Dr.Wael Deeb]

Thanks Lucas ; but I would also be happy to follow Fabien's work and progress until Senpai 2.0 Fabien, do you think you will publish your work in a public version control system ?

A big part of the fun of developing a chess engine is to prepare a big "coup" by releasing an engine hundreds of Elo points stronger than its previous version... The surprise element is very fun... And this is something that such a public development system prevents. Therefore, not all engines author want to go the Stockfish way.

Totally agreed my friend....
Dr.D

[ZirconiumX]

Thanks Lucas ; but I would also be happy to follow Fabien's work and progress until Senpai 2.0 Fabien, do you think you will publish your work in a public version control system ?

A big part of the fun of developing a chess engine is to prepare a big "coup" by releasing an engine hundreds of Elo points stronger than its previous version... The surprise element is very fun... And this is something that such a public development system prevents. Therefore, not all engines author want to go the Stockfish way.

This is part of the reason, though another part, from bits of correspondence with Fabien is that the dev version has a huge amount of debugging framework in it, along with toys like being able to play a game using CPU time inside a single executable that make things slightly difficult to read.

Matthew:out

[Xann]

Fabien, do you think you will publish your work in a public version control system ?

No, sorry about that.

[Marek Soszynski]

Thank you Fabien!

Could you also provide a non-SSE Linux compile? (You already do provide a non-SSE Windows compile.) I did try to compile one myself, but without success.

Incidentally, the non-SSE Windows compile runs okay on Linux via Wine, and sure enough it exhibits the 50% hash bug.

[phenri]

Thank you Fabien!

Could you also provide a non-SSE Linux compile? (You already do provide a non-SSE Windows compile.) I did try to compile one myself, but without success.

Incidentally, the non-SSE Windows compile runs okay on Linux via Wine, and sure enough it exhibits the 50% hash bug.

try those below:
http://www.solidfiles.com/d/a8052b4b89/

Also, if you gave the model name of your CPU, we could try to compile a version close to your machine.

[lucasart]

Thank you Fabien!

Could you also provide a non-SSE Linux compile? (You already do provide a non-SSE Windows compile.) I did try to compile one myself, but without success.

Try this:

Code: Select all

g++ -static-libstdc++ -std=c++11 -pthread -O3 -sse2 -fno-rtti -s ./senpai_10.cpp -o ./senpai_1.0

Obviously you need a recent enough g++ installed that supports c++11.

I doubt anybody uses a "non SSE" machine these days:

Streaming SIMD Extensions (SSE) is a SIMD instruction set extension to the x86 architecture, designed by Intel and introduced in 1999 in their Pentium III series processors

1999 was 15 years ago. It's like the paleolithic

What you probably want is SSE 3, or maybe SSE 2, because your CPU doesn't have SSE 4.2 (essentially hardware popcount).

[Marek Soszynski]

Thank you Fabien!

Could you also provide a non-SSE Linux compile? (You already do provide a non-SSE Windows compile.) I did try to compile one myself, but without success.

Incidentally, the non-SSE Windows compile runs okay on Linux via Wine, and sure enough it exhibits the 50% hash bug.

try those below:
http://www.solidfiles.com/d/a8052b4b89/

Also, if you gave the model name of your CPU, we could try to compile a version close to your machine.

Thank you hugely.
The sse3 compile therein works very well on my machine, and now I am more than happy.

[Marek Soszynski]

What you probably want is SSE 3, or maybe SSE 2, because your CPU doesn't have SSE 4.2 (essentially hardware popcount).

Quite. And now an appropriate compile has been provided.