Barbarossa: new release 0.6.0

[nionita]

Hi all,

I just release a new version of Barbarossa (see https://github.com/nionita/Barbarossa/r ... /tag/0.6.0).

It is rated about 90 Elo in self play over the previous version 0.5.0.

Changes from the previous version:
- Compilation with more recent GHC / Stack (GHC 8.8.4 / stack LTS-16.20)
- New eval terms: weak pawns, optimized
- Eval parameters & weights structures generated with Template Haskell
- Fixed extended futility pruning up to depth 6
- Simplification of time management and helper threads
- Better inlining and strictness in code
- Search depth 1 normally (no full window)
- SelfPlay: add new parameter for nodes budget
- Search time reduction for first move
- Use only one CPU for all threads (search + helper)
- Two abort times, unlimited extensions
- Killer list with 3 killers
- Better history scale when maximum reached
- Higher value for king safety
- Better speed, especially in QS

I provided binaries for Windows 10 and Linux (both SSE4.2 & 64 bit).

I would appreciate if the CCRL testers will take this new version for tests with other engines.

Have a nice weekend,
Nicu

[Guenther]

Hi all,

I just release a new version of Barbarossa (see https://github.com/nionita/Barbarossa/r ... /tag/0.6.0).

It is rated about 90 Elo in self play over the previous version 0.5.0.

Changes from the previous version:
- Compilation with more recent GHC / Stack (GHC 8.8.4 / stack LTS-16.20)
- New eval terms: weak pawns, optimized
- Eval parameters & weights structures generated with Template Haskell
- Fixed extended futility pruning up to depth 6
- Simplification of time management and helper threads
- Better inlining and strictness in code
- Search depth 1 normally (no full window)
- SelfPlay: add new parameter for nodes budget
- Search time reduction for first move
- Use only one CPU for all threads (search + helper)
- Two abort times, unlimited extensions
- Killer list with 3 killers
- Better history scale when maximum reached
- Higher value for king safety
- Better speed, especially in QS

I provided binaries for Windows 10 and Linux (both SSE4.2 & 64 bit).

I would appreciate if the CCRL testers will take this new version for tests with other engines.

Have a nice weekend,
Nicu

Thanks for the release Nicu!
Would you mind to provide a non-popcount binary too (may be for SSSE3, which would cover most old hardware still used by people at talkchess)?

[nionita]

Hi Guenther,

I will try to make a non-popcount version, the problem is, I cannot test it, as I have no hardware without SSE4.2. I don't think the functionality will be broken, but the the speed could be very different, as I already saw in previous releases..

[nionita]

I will try to make a non-popcount version, the problem is, I cannot test it, as I have no hardware without SSE4.2. I don't think the functionality will be broken, but the the speed could be very different, as I already saw in previous releases..

I uploaded a Windows 10 version without popcount (a bit slower, 5-10%) - I hope it works well, I could not test on other machines.
Nicu

[Guenther]

I will try to make a non-popcount version, the problem is, I cannot test it, as I have no hardware without SSE4.2. I don't think the functionality will be broken, but the the speed could be very different, as I already saw in previous releases..

I uploaded a Windows 10 version without popcount (a bit slower, 5-10%) - I hope it works well, I could not test on other machines.
Nicu

Thanks that works :)