Post all test results for Rybka 2.3.1 Beta in this thread.
Inspired by FWCC's post (:)) where he said:
Inspired by FWCC's post (:)) where he said:
>Instead of searching through a LOT of threads and posts I thought it would be easier to have a thread started that would contain the test results. I am very interested in the test results to see if [this version] is indeed strongest of them all and also for general interest in computer chess to the public. Please feel free to also post ANY results including latest CEGT results too, or of course any personal tests results that you may have performed. Thanks a million!
Thank you Dadi,I kinda started a thread that says"2.3.1 test/impressions allready.But hey, the more the better.
FWCC
FWCC
Two engine tournaments on AMD Athlon 64 processor 3200+, 1.99 GHz, 512 MB RAM, WinXP SP2, 128 MB hash, ponder off, 3-4-5 and some (+/- 130 Gig) 6-men TBs:
200 games where 2.2n2 has Constant Time Usage, 2.3.1 has Varied Time Usage:
Blitz:1'+1"
1 Rybka 2.3.1 32-bit 2400 +16/-5/=76 55.67 54.0/97 = 55.7%
2 Rybka 2.2n2 32-bit 2400 +5/-16/=76 44.33 43.0/97 = 44.3%
250 games with both engines on Varied Tme Usage:
Blitz:1'+1"
1 Rybka 2.3.1 32-bit 2400 +14/-5/=92 54.05 60.0/111 = 54%
2 Rybka 2.2n2 32-bit 2400 +5/-14/=92 45.95 51.0/111 = 46%
Though the number of games is low, and the difference insignificant, it's probably correct that Time Usage has an slight influence on the playing strength.
200 games where 2.2n2 has Constant Time Usage, 2.3.1 has Varied Time Usage:
Blitz:1'+1"
1 Rybka 2.3.1 32-bit 2400 +16/-5/=76 55.67 54.0/97 = 55.7%
2 Rybka 2.2n2 32-bit 2400 +5/-16/=76 44.33 43.0/97 = 44.3%
250 games with both engines on Varied Tme Usage:
Blitz:1'+1"
1 Rybka 2.3.1 32-bit 2400 +14/-5/=92 54.05 60.0/111 = 54%
2 Rybka 2.2n2 32-bit 2400 +5/-14/=92 45.95 51.0/111 = 46%
Though the number of games is low, and the difference insignificant, it's probably correct that Time Usage has an slight influence on the playing strength.
Engine Match Rybka 2.3.1 vs Rybka 2.2n2 & Rybka 2.3lk, 64 games (32 + 32)
256MB hash tables, 3-4-5 TB, ponder off
Hardware: Dual Core AMD Opteron(tm) Processor 165 2400 MHz with 1.024 MB Memory
Operating system: Microsoft Windows XP 64 Bit Professional Service Pack 1 (Build 3790)
LEVEL: Tournament 40 moves / 10 min.
Results:
Rybkav2.3.1.mp.x64 - Rybkav2.2n2.mp.x64 : 18,0/32 10-6-16 (=====01==11=1=01=01=1=1=100==01=) 56% +42
Rybkav2.3.1.mp.x64 - Rybkav2.3.lk.mp.x64 : 15,5/32 6-7-19 (==0==0=1000====0==11===0=111====) 48% -14
Cheers,
256MB hash tables, 3-4-5 TB, ponder off
Hardware: Dual Core AMD Opteron(tm) Processor 165 2400 MHz with 1.024 MB Memory
Operating system: Microsoft Windows XP 64 Bit Professional Service Pack 1 (Build 3790)
LEVEL: Tournament 40 moves / 10 min.
Results:
Rybkav2.3.1.mp.x64 - Rybkav2.2n2.mp.x64 : 18,0/32 10-6-16 (=====01==11=1=01=01=1=1=100==01=) 56% +42
Rybkav2.3.1.mp.x64 - Rybkav2.3.lk.mp.x64 : 15,5/32 6-7-19 (==0==0=1000====0==11===0=111====) 48% -14
Cheers,
Arena 1.1, 40/10min, Nalimov 3-4-5
book: Mainbook.abk (H.Schnapp) -only 6 ply-
Ponder Off, Hash: 64Mb
Pentium4 3.06GHz, 1Gb; WinXP SP2
book: Mainbook.abk (H.Schnapp) -only 6 ply-
Ponder Off, Hash: 64Mb
Pentium4 3.06GHz, 1Gb; WinXP SP2
Motor Puntaje Ry 2.3.1 Ry 2.3 Ry 2.2n2 S-B
1: Rybka 2.3.1 20.0/32 ················ 1====111=1110=== 0===1011===011== 283,00
2: Rybka 2.3 15.5/32 0====000=0001=== ················ =======1==011111 231,25
3: Rybka 2.2n2 12.5/32 1===0100===100== =======0==100000 ················ 225,25
So then 2.3.1 looks to be a monster.I say congratulations Vas on a job well done.After the bugs are worked out this engine will have RECORD ratings/Mad ratings that will increase Vas's lead in this field for sure!
FWCC
FWCC
On a Quadcore QX6700 running WinXP 64bit and 5 0 time control my test was against 2.1d3 (the favourite Playchess blitz engine). Results confirm that Rybka 2.3.1 is the strongest Rybka yet .... +13 -2 = 35
Only 50 games but it seems that it is siginificantly stronger.
Only 50 games but it seems that it is siginificantly stronger.
Could someone re-run these 4 preset openings for both sides with very fast hardware please? Curious bout the result then. Someone with MP versions and same timecontrols (40/12min, 40/12 min, 40/15 min)
Openings are quite interesting (see attachment)
Rybka 2.3.1 at default settings, Hiarcs 11.1 default settings, except Hyper Modern disabled (as it weakens the engine imo)
Hiarcs11.1UCI [edit] - Rybka v2.3.1.w32 : 4,5/8 4-3-1 (01=11100) 56%
-----------------Rybka v2.3.1.w32-----------------
Rybka v2.3.1.w32 - Hiarcs11.1UCI [edit] : 3,5/8 3-4-1 (10=00011) 44%
Greetings,
masomusic
Openings are quite interesting (see attachment)
Rybka 2.3.1 at default settings, Hiarcs 11.1 default settings, except Hyper Modern disabled (as it weakens the engine imo)
Hiarcs11.1UCI [edit] - Rybka v2.3.1.w32 : 4,5/8 4-3-1 (01=11100) 56%
-----------------Rybka v2.3.1.w32-----------------
Rybka v2.3.1.w32 - Hiarcs11.1UCI [edit] : 3,5/8 3-4-1 (10=00011) 44%
Greetings,
masomusic
Attachment: Rybka2.3.1vsHiarcs11.1.zip (27k)
Fritz Interface, Windows XP, Athlon 3200+ single processor, 2 Gb RAM, 512 MB hash for each engine, 64MB for Nalimov 3-4-5-6(some) tablebases. Rykba II openings book.
40/10, 40/10, etc.
Rybka 2.3.1 32-bit ½½½½11½½½01½ 7.0/12
Rybka 2.2n2 32-bit ½½½½00½½½10½ 5.0/12
40/10, 40/10, etc.
Rybka 2.3.1 32-bit ½½½½11½½½01½ 7.0/12
Rybka 2.2n2 32-bit ½½½½00½½½10½ 5.0/12
Fritz Interface, Windows XP, Athlon 3200+ single processor, 2 Gb RAM, 512 MB hash for each engine, 64MB for Nalimov 3-4-5-6(some) tablebases, Rybka II openings book for both engines.
40/10, 40/10, etc.
1 Rybka 2.3.1 32-bit 0½01½½11½1½11111½½1½½101111½½½0111½1½011 28.0/40
2 Toga II 1.3x4 1½10½½00½0½00000½½0½½010000½½½1000½0½100 12.0/40
Yikes.
40/10, 40/10, etc.
1 Rybka 2.3.1 32-bit 0½01½½11½1½11111½½1½½101111½½½0111½1½011 28.0/40
2 Toga II 1.3x4 1½10½½00½0½00000½½0½½010000½½½1000½0½100 12.0/40
Yikes.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill
