Page 2 of 2

Re: Could someone do a Demolito compile please ?

Posted: Sat Apr 20, 2019 12:40 pm
by Modern Times
Guenther wrote:
Sat Apr 20, 2019 9:37 am
Modern Times wrote:
Sat Apr 20, 2019 4:54 am
Tirsa's compile was used to test it here:

http://ccrl.chessdom.com/ccrl/404FRC/
It seems Tirsa did not announce it openly in the CCRL forum.
Will this compilation be available for all users?

Sure, I'll post it in the CCRL Public forum.

http://kirill-kryukov.com/chess/discuss ... =7&t=11072

Re: Could someone do a Demolito compile please ?

Posted: Sat Apr 20, 2019 6:34 pm
by Guenther
Modern Times wrote:
Sat Apr 20, 2019 12:40 pm
Guenther wrote:
Sat Apr 20, 2019 9:37 am
Modern Times wrote:
Sat Apr 20, 2019 4:54 am
Tirsa's compile was used to test it here:

http://ccrl.chessdom.com/ccrl/404FRC/
It seems Tirsa did not announce it openly in the CCRL forum.
Will this compilation be available for all users?

Sure, I'll post it in the CCRL Public forum.

http://kirill-kryukov.com/chess/discuss ... =7&t=11072
Thanks and happy easter!

Re: Could someone do a Demolito compile please ?

Posted: Sun Apr 21, 2019 1:51 am
by lucasart
Modern Times wrote:
Sat Apr 20, 2019 4:54 am
Tirsa's compile was used to test it here:

http://ccrl.chessdom.com/ccrl/404FRC/
Thanks for testing Demolito. I'm a surprised to see it so high in the ranking. Yet, the elo gap to SF is monstruous, which indicates that too few engines support Chess960. It's a shame, because I like 960. It's a bit weird, admittedly, but drastically reduces the draw rate, and measures engines more holistically (ie. you don't start 20 moves out of book in advanced middle game, but you start awkwardly underdevelopped, and have to find a strategy to develop by thinking out of the box, rather than repeating the same patterns).

Re: Could someone do a Demolito compile please ?

Posted: Sun Apr 21, 2019 5:29 pm
by Guenther
Guenther wrote:
Sat Apr 20, 2019 6:34 pm
Modern Times wrote:
Sat Apr 20, 2019 12:40 pm
Guenther wrote:
Sat Apr 20, 2019 9:37 am
Modern Times wrote:
Sat Apr 20, 2019 4:54 am
Tirsa's compile was used to test it here:

http://ccrl.chessdom.com/ccrl/404FRC/
It seems Tirsa did not announce it openly in the CCRL forum.
Will this compilation be available for all users?

Sure, I'll post it in the CCRL Public forum.

http://kirill-kryukov.com/chess/discuss ... =7&t=11072
Thanks and happy easter!
BTW - would Tirsa be able to compile a no popcount version too, as he did in the past?

Re: Could someone do a Demolito compile please ?

Posted: Sun Apr 21, 2019 11:32 pm
by tpoppins

Re: Could someone do a Demolito compile please ?

Posted: Wed Apr 24, 2019 11:19 pm
by lucasart
tpoppins wrote:
Sun Apr 21, 2019 11:32 pm
He would! :>

demolito-20190206_sse41-tp.zip
Thanks to erbenzaehler and AppVeyor, Demolito now has automatically generated Windows compiles with each commit:
* go here: https://github.com/lucasart/Demolito/commits/master
* select the commit you want (eg. the latest), and click the green check mark, then click Details...
* this brings you to an AppVeyor page, click Artifacts

And there you have 3 clang compiles. As always, PEXT is the first choice (fastest but only works on modern computers), if that fails, try POPCNT, and if that fails, try NO_POPCNT.

I'm not 100% sure these compiles really work on Windows machine without additional DLL. Let me know.

Re: Could someone do a Demolito compile please ?

Posted: Wed Apr 24, 2019 11:29 pm
by tpoppins
lucasart wrote:
Wed Apr 24, 2019 11:19 pm
I'm not 100% sure these compiles really work on Windows machine without additional DLL. Let me know.
They do, good job, Lucas.

Re: Could someone do a Demolito compile please ?

Posted: Thu Apr 25, 2019 9:37 am
by Guenther
lucasart wrote:
Wed Apr 24, 2019 11:19 pm
tpoppins wrote:
Sun Apr 21, 2019 11:32 pm
He would! :>

demolito-20190206_sse41-tp.zip
Thanks to erbenzaehler and AppVeyor, Demolito now has automatically generated Windows compiles with each commit:
...
I'm not 100% sure these compiles really work on Windows machine without additional DLL. Let me know.
Thanks Lucas! (thanks also to Tirsa for previous no-popcount compilation)
Checked the no-popcount from appveyor and it works here.