IvanHoe compiles testing

[JackStraw]

i7 @ 3.75 / 2 cores ea / ponder off / 512 mb / old RB + TB / F11 / klo openings

Time : 4 mins + 2 sec

Wins were tough to come by .............

Houdini 1.03 .................... 50.5 .......... +11 , -10 , =79

IvanHoe B49jAx64p ............49.5

[robbolito]

AMD Phenom(tm) II X6 1090T Processor6x @ 3.2 GHz 4,096 MB Memory
Windows 7 Home Premium Edition (Build 7600)
Fritz Benchmark:
Speed: 23.33
KNS: 11196
GUI: CB Rybka 3
Book: Rybka 4- 10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

Code: Select all

1   Houdini 1.03a x64 POPCNT 8_CP  3190  +16/-16/=48 50.00%   40.0/80  1600.00
2   Ivanhoe B49jAx64p              3190  +16/-16/=48 50.00%   40.0/80  1600.00

[robbolito]

Intel(R) Core(TM)2 Quad Q9550 2.83GHz4x @3.5 GHz 4,096 MB Memory
Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)
Fritz Benchmark:
Speed: 20.90
KNS: 10032
GUI: CB Rybka 3
Book: Balanced 16-10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

Code: Select all

1   Ivanhoe B49jAx64         3190  +10/-7/=63 51.88%   41.5/80
2   Houdini 1.03a x64 4_CPU  3190  +7/-10/=63 48.13%   38.5/80

[BB+]

Furthermore, I did a Monte Carlo analysis with the latest IvanHoe:

So this Monte Carlo thing is now working? Seems that Nf3 scored 43.2% and dxe5 was 36.9% which looks quite significant over 5000 trials. How long did it take to get the results [10 ply games are about 5s each I'd guess]? And do the positional evals match what Monte Carlo says? [This was what LK was interested in with Rybka Monte Carlo, especially with openings -- whether the eval differed a lot from the MC result]. I get they are both around -0.4 or so at depth 24. Is there a centipawn-to-win% conversion for IvanHoe, or will that for Rybka work just as well?

[BB+]

The Rybka conversion table by George Tsavdaris is in the thread http://rybkaforum.net/cgi-bin/rybkaforu ... l?tid=6012

Code: Select all

Rybka's eval | % expected score
       0.01         50.43 %
       0.10         54.3 %   
       0.20         58.5 %   
       0.30         62.7 %    
       0.40         66.6 %   
       0.50         70.3 %    
       0.80         79.9 %    
       0.90         82.6 %    
       1.00         84.9 %    
       1.10         87.0 %     
       1.20         88.8 %    
       1.30         90.4 %    
       1.40         91.8 %    
       1.50         93.0 %    
       1.60         94.1 %    

Tables are based on the assumptions:
•Rybka's evaluation score is proportional to the ELO rating difference.
•It is true that 3 ELO difference corresponds to 0.01 evaluation score of Rybka.

VR says: "[...] based on 8 ply monte carlo data [...] roughly we can say that each centipawn of Rybka eval is worth 3 Elo points on average. My earlier work on opening evals of the different versions used 2.5 Elo per centipawn, but this value was pretty much just based on the opening position, so I would have much more confidence in the 3 Elo figure."

This would mean that -0.4 would be -120 Elo, or almost exactly 1/3 of points (33.4%). Neither Nf3 (43.2%) nor dxe5 (36.9%) does that quite bad. However, a small flux in the -0.4 number can change the conclusion slightly (-0.35 would be 35.3%, so 5 centipawns is nearly 2% in score), as could changing 3 to 2.5 in the conversion. But as I say, to me the interest is not so much with a conversion, but when standard search and Monte Carlo show disparate results.

[robbolito]

Intel(R) Core(TM) i5 Q- 750 @2.67 GHz 6.00 GB Memory
Windows 7.x64 Professional (Build 7600)
Fritz Benchmarks:
Speed:16.84
KNS: 8082
GUI: CB Rybka 3
Hash: 256
Book: Rybka 4-10 moves
RB and TB: ON
Ponder: OFF
Blitz:10 0

Code: Select all

1   Ivanhoe B49jAx64p  3190  +19/-6/=45 59.29%   41.5/70
2   Deep Rybka 4 x64   3150  +6/-19/=45 40.71%   28.5/70

[robbolito]

AMD Phenom(tm) II X6 1090T Processor 6x @3.895 GHz 4,096 MB Memory Windows XP 64 Bit Professional Service Pack 2 (Build 3790)
Fritz Benchmark:
Speed: 28.33
KNS: 13599
GUI: CB Rybka 3
Book: Rybka 4-10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

Code: Select all

1   Ivanhoe B49jAx64p        3190  +6/-4/=30 52.50%   21.0/40
2   Houdini 1.03a x64 8_CPU  3190  +4/-6/=30 47.50%   19.0/40

[robbolito]

Intel(R) Core(TM) i7 4CPU 960 3.20GHz8x @ 4.005 GHz with 5.99 MB Memory
Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)
Fritz Benchmark:
Speed: 24.98
KNS: 11988
GUI: CB Rybka 3
Book: Balanced 16-10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Blitz:10' 0

Code: Select all

1   Ivanhoe B49jAx64p  3190  +11/-2/=27 61.25%   24.5/40
2   Deep Rybka 4 x64   3190  +2/-11/=27 38.75%   15.5/40

[robbolito]

Intel(R) Core(TM) i7 4CPU 960 3.20GHz8x @ 4.005 GHz with 5.99 MB Memory
Microsoft Windows XP 64 Bit Professional Service Pack 2 (Build 3790)
Fritz Benchmark:
Speed: 24.98
KNS: 11988
GUI: CB Rybka 3
Book: Balanced 16-10 moves
Hash: 256
RB and TB: ON
Ponder: OFF
Qi7- 960 XP.x64 4 GHZ, Blitz:10' 0

Code: Select all

1   Ivanhoe B49jAx64p  3190  +21/-10/=59 56.11%   50.5/90
2   Deep Rybka 4 x64   3190  +10/-21/=59 43.89%   39.5/90

[robbolito]

Intel(R) Core(TM) i5 Q- 750 @2.67 GHz 6.00 GB Memory
Windows 7.x64 Professional (Build 7600)
Fritz Benchmarks:
Speed:16.84
KNS: 8082
GUI: CB Rybka 3
Hash: 256
Book: Rybka 4-10 moves
RB and TB: ON
Ponder: OFF
Blitz:10 0

Code: Select all

1   IvanHoe T49j2.x64P  3190  +9/-8/=23 51.25%   20.5/40
2   Deep Rybka 4 x64    3150  +8/-9/=23 48.75%   19.5/40