Re: New engine releases
Posted: Tue Jan 15, 2019 11:39 am
Xiphos 0.4.18 with tbs support: https://github.com/milostatarevic/xipho ... master/bin
https://github.com/milostatarevic/xiphos
https://github.com/milostatarevic/xiphos
I tried the windows old and ancient binaries on my Win7-64 quadcore, which supports up to SSE 4.1.jorose wrote: ↑Mon Jan 21, 2019 2:53 pm Winter 0.4a is released, change to 0.4 is cosmetic and they are identical in playing strength.
Basically I fixed UCI output of selective depth and updated the version number.
Perhaps more important is that I did not make a native Windows compile, instead it is cross compiled from my regular linux system. I am currently on vacation for a few days and won't be able to confirm it works on a Windows computer on my own until I get back. Until then it would be great if someone could run any of the binaries and confirm they work in a pm, reply to here or message on discord. I considered waiting until I get back to make the release, however I felt this is a good dry run, as I am not expecting to have access to a private Windows system from February until at least August of this year.
In the future I will at least check versions with Wine.
I thought my 0.3 Old compile also expected SSE4.2, so either I remembered that incorrectly or the compiler didn't use any of the SSE4.2 operations despite having access to them. In any case, I am glad the cross compiles seem to work. I have updated the Windows binaries with an "older" compile which expects just SSE4.1. When I get home I'll see if there is a speed difference between them.Guenther wrote: ↑Mon Jan 21, 2019 4:55 pmI tried the windows old and ancient binaries on my Win7-64 quadcore, which supports up to SSE 4.1.jorose wrote: ↑Mon Jan 21, 2019 2:53 pm Winter 0.4a is released, change to 0.4 is cosmetic and they are identical in playing strength.
Basically I fixed UCI output of selective depth and updated the version number.
Perhaps more important is that I did not make a native Windows compile, instead it is cross compiled from my regular linux system. I am currently on vacation for a few days and won't be able to confirm it works on a Windows computer on my own until I get back. Until then it would be great if someone could run any of the binaries and confirm they work in a pm, reply to here or message on discord. I considered waiting until I get back to make the release, however I felt this is a good dry run, as I am not expecting to have access to a private Windows system from February until at least August of this year.
In the future I will at least check versions with Wine.
The ancient binary works here. Interesting though, that with Winter 0.3, even the old version worked.
Well, as long as it runs here I wouldn't even be bothered, if it is called 'pleistocene build'jorose wrote: ↑Mon Jan 21, 2019 7:35 pmI thought my 0.3 Old compile also expected SSE4.2, so either I remembered that incorrectly or the compiler didn't use any of the SSE4.2 operations despite having access to them. In any case, I am glad the cross compiles seem to work. I have updated the Windows binaries with an "older" compile which expects just SSE4.1. When I get home I'll see if there is a speed difference between them.
Also I still need to come up with a nice way to make people not be embarassed about which compile they need to use, but understand which ones they should try. Having your machine called ancient is probably not too nice, yet my impression is, that it is the most popular compile
The ancient compile works fine on my PC, also no problems with parallel games. Here a 4'+2" tournament:jorose wrote: ↑Mon Jan 21, 2019 2:53 pm Winter 0.4a is released, change to 0.4 is cosmetic and they are identical in playing strength.
Basically I fixed UCI output of selective depth and updated the version number.
Perhaps more important is that I did not make a native Windows compile, instead it is cross compiled from my regular linux system. I am currently on vacation for a few days and won't be able to confirm it works on a Windows computer on my own until I get back. Until then it would be great if someone could run any of the binaries and confirm they work in a pm, reply to here or message on discord. I considered waiting until I get back to make the release, however I felt this is a good dry run, as I am not expecting to have access to a private Windows system from February until at least August of this year.
In the future I will at least check versions with Wine.
Code: Select all
Score (Tie) Pa Wi Wy Bu To Fr De Ka Ch Sh Co Rh Fr
----------------------------------------------------------------------------------
1: Pawny 1.2 16.5 / 24 184.00 XX == =0 == =1 =0 11 11 01 01 11 11 11 (+13 -4 =7)
2: Winter 0.4a 14.0 / 24 168.50 == XX =1 11 10 == 01 =0 01 =1 01 11 0= (+10 -6 =8)
3: WyldChess v1.51 14.0 / 24 161.50 =1 =0 XX 0= 10 == 0= 01 11 11 1= 01 1= (+10 -6 =8)
4: Bugchess2 1.9 13.5 / 24 157.00 == 00 1= XX 00 11 11 =0 =1 1= =1 10 01 (+10 -7 =7)
5: TomitankChess 3.0 13.0 / 24 150.25 =0 01 01 11 XX 0= == 10 1= 0= 0= =1 11 (+9 -7 =8)
6: Francesca MAD 0.21 12.0 / 24 144.25 =1 == == 00 1= XX 00 1= 0= 10 1= 11 =0 (+7 -7 =10)
7: Devel 2.0000 12.0 / 24 137.50 00 10 1= 00 == 11 XX 01 =0 01 11 01 10 (+10 -10 =4)
8: Karballo 1.8 11.0 / 24 130.25 00 =1 10 =1 01 0= 10 XX 10 == 00 01 =1 (+8 -10 =6)
9: Chess Tiger 2007.1 11.0 / 24 130.25 10 10 00 =0 0= 1= =1 01 XX =1 10 0= == (+7 -9 =8)
10: Shield 2.1 11.0 / 24 126.25 10 =0 00 0= 1= 01 10 == =0 XX == 01 11 (+7 -9 =8)
11: Colossus 2008b 10.0 / 24 114.25 00 10 0= =0 1= 0= 00 11 01 == XX =0 =1 (+6 -10 =8)
12: Rhetoric 1.4.3 10.0 / 24 111.50 00 00 10 01 =0 00 10 10 1= 10 =1 XX 1= (+8 -12 =4)
13: Fridolin 3.00 8.0 / 24 98.00 00 1= 0= 10 00 =1 01 =0 == 00 =0 0= XX (+4 -12 =8)
----------------------------------------------------------------------------------
156 games: +55 -54 =47