New Berserk soon?

[Jouni]

I am tired of SF versions. No significant progress in one year. In TCEC Berserk is leading over Dragon 3.3.

[mclane]

Number one changes. From time to time.

[Graham Banks]

Jay has said sometime in 2024.

[peter]

I am tired of SF versions. No significant progress in one year. In TCEC Berserk is leading over Dragon 3.3.

Did you notice in this Tiebreak Type Test one engine always has 10x as much time than the other one? Like that the bookless games shall get full points at all.
Again I didn't realize that at once by myself neither but after 100 games side's have changed and now it's Dragon with the full 10'+1" that Berserk had before and after 53 games now Dragon's leading by 3 points. Berserk in the 100 before won by 4, regards

[peter]

after 53 games now Dragon's leading by 3 points. Berserk in the 100 before won by 4

After the full 100 it was 56:44 for Dragon with the time advantage on his side, regards

[Ciekce]

I am tired of SF versions. No significant progress in one year.

absolutely clueless

[Jouni]

Thanks to Jim Ablett we have a new version . I ran fast test 60+0,6 UHO book:

Code: Select all

Score of berserk_20240311 vs berserk12: 67 - 62 - 171 [0.508]
...      berserk_20240311 playing White: 64 - 2 - 84  [0.707] 150
...      berserk_20240311 playing Black: 3 - 60 - 87  [0.310] 150
...      White vs Black: 124 - 5 - 171  [0.698] 300
Elo difference: 5.8 +/- 25.8, LOS: 67.0 %, DrawRatio: 57.0 %
300 of 300 games finished.

Nothing special. But note Jim's compile seems to be basic 64 compile. No AVX2/BMI so 10-20% slower than Berserk12.

[Graham Banks]

Thanks to Jim Ablett we have a new version

Unofficial release though.

[Modern Times]

I'd call it a development version. The code is official.

[dkappe]

I am tired of SF versions. No significant progress in one year. In TCEC Berserk is leading over Dragon 3.3.

Based on our testing (Torch), SF has made substantial progress. Not sure where this no-progress-in-SF meme comes from. We‘ll catch SF16, but SF16.1 is a tough nut to crack.