cheng3 uci v1.02 (now including source)

[mar]

Hi all, I'm releasing the (messy) sources under GPL(v3).
The zip includes binary as well (no strength change from 1.01).
The source includes VS2008 solution and project files.
The link is here: http://www.mediafire.com/?9ap47cl551bkxoe

Regards
Martin

[UncombedCoconut]

Awesome. I hadn't expected the engine to be SMP-capable, let alone to see nearly complete support for Unix-like operating systems. I'll need to find some time to play with it.
(The last time I offered to help port an open-source SMP engine to Linux ended in failure, so I won't do that again. But let me know if you'd like help testing anything.)

[Dann Corbit]

Here is a version with a 64 bit build:
http://cap.connx.com/chess-engines/new- ... ng3_uci.7z

There is a bug in the original where array bounds are over-written.

[tmokonen]

Thank you, Martin and Dann.

[mar]

Awesome. I hadn't expected the engine to be SMP-capable, let alone to see nearly complete support for Unix-like operating systems. I'll need to find some time to play with it.
(The last time I offered to help port an open-source SMP engine to Linux ended in failure, so I won't do that again. But let me know if you'd like help testing anything.)

It's not SMP capable Justin. It just runs the main search in a separate thread.

[UncombedCoconut]

That actually makes a lot more sense.

[mar]

Here is a version with a 64 bit build:
http://cap.connx.com/chess-engines/new- ... ng3_uci.7z

There is a bug in the original where array bounds are over-written.

Wow thanks Dann it seems you changed a lot (WinMerge shows tons of differences, perhaps formatting/whitespace?), I suppose it was for the sake of compiler compatibility + adding enum names and consts where appropriate.
Can you tell me where the array indexing bug is to be found?

Second: the engine is optimized for 32-bit so I don't really think the 64-bit will be any faster (am I wrong? I can't test it for I don't have access to 64-bit OS). I'm 100% sure there's a way to use bsf/bsr in the x64 version so that things go faster (unfortunately __asm doesn't work in MS compiler in x64 mode). The multi-table-lookups used in non-MSC compilers (getLSB/MSB) will be slower IMO.

[mar]

core/microcore already contains preliminary support for pthreads but the Events are still incomplete (condition variables should do). I released in a hurry...

[mar]

Oh I see you already did it using intrinsics So it should be faster. Thanks again!

[mar]

... by which I mean it should be possible to port to linux (though I don't know if gcc will eat msc code