Update of latest IvanHoe compiles.

[Tomcass]

UPDATED: SUMMARY OF TESTS WITH LATEST IVANHOE COMPILES. 4.311 GAMES AND 22 COMPILES.

Computer: i7 975 4 real cores + 4 Hyper Thread.
Fritz Benchmark: Speed 26,73 KNS 12.828
Gui: Fritz 12
Time Control: 2' + 1
Book: H.S. Masterbook 3.0
Ponder: Off.
No Robbo, Triple or Nalimov.
No crashes.


Latest tests:

201012Ivanhoe48vA 2010

Ivanhoe B48vAx64a (x4) - Deep Rybka 4 SSE42 x64(x4) 54.5 - 45.5 +21/=67/-12 54.50%
Ivanhoe B48vAx64a (x4) - Houdini 1.03a x64 8_CPU (x4) 53.5 - 46.5 +22/=63/-15 53.50%

201012Ivanhoe47vAPPan 2010

Ivanhoe B47cAx64 (x4) - Deep Rybka 4 SSE42 x64(x4) 53.5 - 46.5 +17/=73/-10 53.50%
Ivanhoe B47cAx64 (x4) - Houdini 1.03a x64 8_CPU (x4) 52.0 - 48.0 +17/=70/-13 52.00%

201012Ivanhoe47KLO 2010

Ivanhoe B47cAx64(x4)KLO - Deep Rybka 4 x64 (x4) 55.5 - 44.5 +24/=63/-13 55.50%
Ivanhoe B47cAx64(x4)KLO - Houdini 1.03a x64 8_CPU (x4) 52.5 - 47.5 +17/=71/-12 52.50%

UPDATED RANKING AFTER 4.311 GAMES

AGAINST DEEP RYBKA 4 SSE42

1.- Ivanhoe B50kBx64p 57,00%
2.- IvanhoeB50tAp 56.50%
3.- Ivanhoe B47cAx64(x4)KLO 55.50%
4.- IvanHoe 9.49b x64PopCNT_KLO 55,0%
5.- Ivanhoe B48vAx64a 54,50%
6.- Ivanhoe B47cAx64Ppan 53,50%
7.- IvanHoe 9.50b x64 52,50%
8/9.- IvanHoe T0.5.4.1 52.00%
8/9.- IvanHoe B50hBp 52.00%
10.- IvanHoe T50K 51.00%
11/12- IvanHoe B50kAx64p 50,50%
11/12.- Ivanhoe B49jAx64p 50,50%
13.- Ivanhoe B49fAx64p 50,00%
14.- IvanHoe T0.7 48,50%
15.- IvanHoe 9.50b x64KLO 48.21%
16.- IvanHoe T0.5.4.2 48,00%
17.- IvanHoe T49j 47,50%
18.- IvanHoe T0.5.1POPCNT 46.50%
19.- Ivanhoe-BetaWH_87j_w64 45,00%
20.- IvanHoe T0.5.2POPCNT 44.00%
21.- IvanhoeB50kB 43,50%
22.- IvanHoe T0.5.3 40.50%

On average, the tested compiles got a 50.10% against DR4.

AGAINST HOUDINI 1.03a

1.- IvanhoeB50kB 58.00%
2.- IvanHoe 9.49b x64PopCNT_KLO 55.00%
3.- Ivanhoe B49fAx64p 54,50%
4/5/6 .- IvanHoe T0.5.1POPCNT 53.50%
4/5/6.- Ivanhoe B50kBx64p 53,50%
4/5/6.- Ivanhoe B48vAx64a 53,50%
7/8/9.- IvanHoe T0.5.4.1 53.00%
7/8/9.- IvanHoe B50hBx64p(x4) 53.00%
7/8/9.- IvanHoe T0.7 53,00%
10/11/12 .- IvanhoeB50tAX64p 52.50%
10/11/12.- IvanHoe 9.50b x64 52,50%
10/11/12.- Ivanhoe B47cAx64KLO 52,50%
13.- Ivanhoe B47cAx64 PPan 52.00%
14/15.- IvanHoe T50K(x8) 51.50%
14/15.- Ivanhoe B49jAx64p 51.50%
16.- IvanHoe T0.5.3 51.00%
17/18.- IvanHoe T0.5.4.2Popcnt 50.00%
17/18.- IvanHoe T49j 50.00%
19.- Ivanhoe-BetaWH_87j_w64 49,50%
20.- IvanHoe B50kAx64p(x8) 49.00%
21.- IvanHoe T0.5.2POPCNT (x8) 48.00%
22.- IvanHoe 9.50b x64KLO(x8) 47,27%

On average, the tested compiles got a 52.01% against Houdini 1.03a

GLOBAL SCORE AFTER 4.311 GAMES

1.- Ivanhoe B50kBx64p Peterpan 55.25%
2.- IvanHoe 9.49b x64PopCNT_KLO 55.00%
3.- IvanhoeB50tAX64p 54.50%
4.- Ivanhoe B48vAx64a 54,00%
5.- Ivanhoe B47cA KLO 54,00%
6.- Ivanhoe B47cA Peterpan 52,75%
7/8/9.- IvanHoe B50hBx64p 52.50%
7/8/9.- IvanHoe T0.5.4.1 52.50%
7/8/9.- IvanHoe 9.50b x64 52,50%
10.- Ivanhoe B49fAx64p 52,25%
11.- IvanHoe T50K 51,25%
12.- Ivanhoe B49jAx64p 51.00%
13.- IvanhoeB50kB 50.75%
14.-IvanHoe T0.7 50,75%
15.- IvanHoe T0.5.1POPCNT 50,00%
16.- IvanHoe B50kAx64p 49,75%
17.- IvanHoe T0.5.4.2 49,00
18.- IvanHoe T49j 48.75%
19.- IvanHoe 9.50b x64KLO(x8) 47,74%
20.- Ivanhoe-BetaWH_87j_w64 47,25%
21.- IvanHoe T0.5.2POPCNT (x8) 46,00%
22.- IvanHoe T0.5.3 45,75%

On average, the tested compiles got a 51.06% against both engines. Only one month ago this percentage was 50.66%. Latest versions (48 and 47) seem to be slighly better than the previous ones.

Best regards from Barcelona.

Tom

[ernest]

UPDATED: SUMMARY OF TESTS WITH LATEST IVANHOE COMPILES

Sorry, Tom, I also have done tests of your "better" IvanHoes against Deep Rybka 4: and I must say that they can only be considered as being equivalent in strength to DR4, not better (never confirmed your 57%, 55% data after several 100's of games)
... which is not the case of Houdini 1.5, clearly better!

[Tomcass]

Hello, Ernest.

As usually I say, I do not pretend to state which is the best program. I test only for my own enjoyment and share my results with chess community.

In my tests the results were the reported ones. I know that 200 games per compile are not enough to say that one program is better than other. They are only an indicator. The Time Control is also essential to understand the results and so is the speed of the computer and the openings book used. So our results are not necessarily contradictory.

My opinion -I can be obviously wrong- is that some IvanHoe compiles, at this time control, with this openings book and working in my computer -12,600 kns Fritz benchmark- are stronger than DR4.

Regards,

Tom.