Page 1 of 1
Asmfish further development
Posted: Mon Oct 01, 2018 10:43 am
by Guenther
For the people who didn't realize yet that Asmfish compilations with newer SFdev commits are still being developed.
(I happened to know this myself just since a couple of days ago - the maintainer is Justin Dehorty, aka counterply)
The latest version is from 09-29-2018 - the previous Lantonov versions were stuck at July.
(The page header wrongly says 'Latest version: 9/24/18...', obviously the header is manually updated and was forgotten to be changed )
https://github.com/Counterply/asmFish/t ... CounterPly
Re: Asmfish further development
Posted: Mon Oct 01, 2018 4:20 pm
by zullil
Guenther wrote: ↑Mon Oct 01, 2018 10:43 am
For the people who didn't realize yet that Asmfish compilations with newer SFdev commits are still being developed.
(I happened to know this myself just since a couple of days ago - the maintainer is Justin Dehorty, aka counterply)
The latest version is from 09-29-2018 - the previous Lantonov versions were stuck at July.
(The page header wrongly says 'Latest version: 9/24/18...', obviously the header is manually updated and was forgotten to be changed )
https://github.com/Counterply/asmFish/t ... CounterPly
Thanks! I had no idea asmfish was still being actively developed.
Re: Asmfish further development
Posted: Mon Oct 01, 2018 7:48 pm
by ernest
Guenther wrote: ↑Mon Oct 01, 2018 10:43 am
The latest version is from 09-29-2018 - the previous Lantonov versions were stuck at July.
(The page header wrongly says 'Latest version: 9/24/18...', obviously the header is manually updated and was forgotten to be changed )
https://github.com/Counterply/asmFish/t ... CounterPly
Hi Guenther,
Does this latest version correspond to a given SF_dev ?
Re: Asmfish further development
Posted: Mon Oct 01, 2018 8:37 pm
by Guenther
ernest wrote: ↑Mon Oct 01, 2018 7:48 pm
Guenther wrote: ↑Mon Oct 01, 2018 10:43 am
The latest version is from 09-29-2018 - the previous Lantonov versions were stuck at July.
(The page header wrongly says 'Latest version: 9/24/18...', obviously the header is manually updated and was forgotten to be changed )
https://github.com/Counterply/asmFish/t ... CounterPly
Hi Guenther,
Does this latest version correspond to a given SF_dev ?
Good question. I automatically assumed latest patches (4) would have been the (4) from 09-27 (except one for typos later the day),
but after cross-checking I realized they are quite behind there.
Anyhow if one compares 'Commits'
https://github.com/Counterply/asmFish/c ... d8e7a3a0cc
Code: Select all
4 more patches bench 4970737
Simplify capture bonus bench 5128885
official-stockfish/Stockfish@231103d
Quiet move soft fail high bonus bench 5021172
official-stockfish/Stockfish@528507b
Optimze contempt value (21) bench 4886199
official-stockfish/Stockfish@86e4675
Simplify capture pruning margin formula bench 4970737
official-stockfish/Stockfish@88de112
we can see that those currently correspond to patches in official SF from June11 and 06!
(compare the hex IDs)
https://github.com/official-stockfish/S ... d1a8efd+69
Another look at the previous (which also was not the first asmfish fork) asmfish by Lantonov
reveals that 'counterply' already maintained that one solely too since 11-04-2018
https://github.com/lantonov/asmFish/com ... 1f0742e+69
Re: Asmfish further development
Posted: Tue Oct 02, 2018 12:32 am
by ernest
Guenther wrote: ↑Mon Oct 01, 2018 8:37 pm
......
Another look at the previous ... asmfish by Lantonov
Yes, Guenther, precisely, Lantonov's latest asmfish corresponds to SF_dev of May 14 (you can check by the name of the patch and, of course, by the bench number).
And the new "further development" is barely some days later.
So, despite this good will, which is welcome, asmfish is desperately lagging many months behind the latest SF_dev.
Re: Asmfish further development
Posted: Tue Oct 02, 2018 3:47 am
by ernest
Besides, the new development only includes 4 patches from May 14 to June 11 (SF_dev has many more for that month interval...).
With the result that this latest asmfish corresponds to no SF_dev version (and its bench number cannot be matched).