Testing Houdini 3.0 Pro

[Tomcass]

Houdini 3

I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

201210Houdini3Pro 2012

Houdini 3 Pro x64_6 - Critter 1.6 64-bitx6_nob 25.0 - 15.0 +14/=22/-4 62.50%
Houdini 3 Pro x64_6 - Deep Rybka 4.1 SSE42 x64 (x6) 29.5 - 10.5 +23/=13/-4 73.75%
Houdini 3 Pro x64_6 - Houdini 2.0c Pro x64(x6) 24.0 - 16.0 +16/=16/-8 60.00%
Houdini 3 Pro x64_6 - Sfish 121014Qi x64 SSE4.2_ 28.0 - 12.0 +19/=18/-3 70.00%

Average Score: 66.56%

160 Games: http://www.mediafire.com/?krmmzngzzi688q8

I7 9750 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Rybka 3
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

Houdini 3 Pro x64_6 - Ivanhoe B50kb PeterPan 63.5 - 36.5 +35/=57/-8 63.50%

100 Games: http://www.mediafire.com/?b0vakfw5sorodpy

Houdini 3.0 Pro is a real monster. Some of its games are really impressive. A true pleasure for the computer chess lovers. Anyway my favourite Ivanhoe B50kb by PeterPan has scored a very decent 36.5% against him. Almost at Critter 1.6a level!.

Regards from Barcelona.

Tom.

[WuShock]

i7 940 @ 2.93
2 threads ea
F12
Ponder off
No egtb
512 MB
klo_openings

Time Control : 4 mins + 2 secs

[/img]

Almost hard to believe........

Looks like Vas will need to add another 100 cores to that cluster............

Tom G.

[Tomcass]

HOUDINI 3 Tactical Mode.

I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

201210Houdini3_TacticalModel

Houdini 3 Pro x64x6TM- Critter 1.6 64-bitx6_nob 24.0 - 16.0 +13/=22/-5 60.00%
Houdini 3 Pro x64x6TM- Deep Rybka 4.1 SSE42 x64(x6 24.5 - 15.5 +13/=23/-4 61.25%
Houdini 3 Pro x64x6TM - Houdini 2.0c Pro x64(x6) 22.5 - 17.5 +14/=17/-9 56.25%
Houdini 3 Pro x64x6TM - Sfish 121014Qi x64 SSE4.2_ 30.0 - 9.0 +25/=10/-4 76.92%

Average Score: 63.60%

159 Games: http://www.mediafire.com/?00hie646snsdn6c

I7 9750 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Rybka 3
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

Houdini 3 Std Tac x64_4 - Ivanhoe B50kb PeterPan 60.5- 39.5 +34/=53/-13 60.5%

100 Games: http://www.mediafire.com/?4v46jau2r6c4te8

Whatching the games my feeling is that it is not so 'tactical' as I expected. Although it is obviously more aggressive than the 'non tactical' mode. Anyway this 'tactical' version of this engine is extremely strong, impressive at the mid and end-game, although not so strong as 'non tactical' against DR 4.1. Again a great result for Ivanhoe B50kb by PeterPan.

Regards,

Tom.

[WuShock]

i7 940 @ 2.93
2 threads ea
F12
Ponder off
No egtb
512 MB
klo_openings

Time Control : 4 mins + 2 secs

[/img]

[Tomcass]

I have repeated the test increasing the TC. Now the test is above 7 ABN (Average Billion Nodes) per game for Houdini 3.0 Pro. The previous ones reached 3,5 ABN. The result is even more impressive.

I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 5 minutes +3 sec/ game.
Ponder: Off
No tablebases
ABN = 7.3

201210Houdini3Pro_LTC 2012

Houdini 3 Pro x64_6 - Critter 1.6 64-bitx6_nob 25.5 - 14.5 +14/=23/-3 63.75%
Houdini 3 Pro x64_6 - Deep Rybka 4.1 SSE42 x64 (x6) 32.5 - 7.5 +26/=13/-1 81.25%
Houdini 3 Pro x64_6 - Houdini 2.0c Pro x64(x6) 27.0 - 13.0 +19/=16/-5 67.50%

Average Score: 70.83% (Simply amazing!!: 59 wins and only 9 losses).

120 Games: http://www.mediafire.com/?wfn9hond3l3bn6o


I7 9750 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Rybka 3
Time control: 5 minutes +3 sec/ game.
Ponder: Off
No tablebases
ABN = 5.2

201210Houdini3.0Pro_LTC 2012

1 Houdini 3 x64Stx4 +34/=60/-6 64.00% 64.0/100
2 Ivanhoe B50kBx64p(x4)BO +6/=60/-34 36.00% 36.0/100

Average Score: 64.00%

100 Games: http://www.mediafire.com/?2disbbobwcg6vwh

Again this PeterPan's old Ivan scored almost at Critter 1.6 level and better than Houdini 2.0. (Please note that the comparison is not totally fair. It uses Rybka 3 book -well tuned against Houdini- and also that the Houdini used is 3.0 Standard, not Pro).

Regards from Barcelona.

Tom.

[WuShock]

i7 940 @ 2.93
2 threads ea
F12
Ponder off
No egtb
512 MB
klo_openings

Time Control : 4 mins + 2 secs

[/img]

[WuShock]

i7 940 @ 2.93
1 thread ea
F12
Ponder off
No egtb
512 MB
klo_openings

Time Control : 4 mins + 2 secs



[/img]

Awaiting Komodo 6 MP ...........

[geots]

Houdini 3

I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

201210Houdini3Pro 2012

Houdini 3 Pro x64_6 - Critter 1.6 64-bitx6_nob 25.0 - 15.0 +14/=22/-4 62.50%
Houdini 3 Pro x64_6 - Deep Rybka 4.1 SSE42 x64 (x6) 29.5 - 10.5 +23/=13/-4 73.75%
Houdini 3 Pro x64_6 - Houdini 2.0c Pro x64(x6) 24.0 - 16.0 +16/=16/-8 60.00%
Houdini 3 Pro x64_6 - Sfish 121014Qi x64 SSE4.2_ 28.0 - 12.0 +19/=18/-3 70.00%

Average Score: 66.56%

160 Games: http://www.mediafire.com/?krmmzngzzi688q8

I7 9750 3.33 Ghz.
4 real cores
GUI: Fritz 12
Book: Rybka 3
Time control: 2 minutes +2 sec/ game.
Ponder: Off
No tablebases

Houdini 3 Pro x64_6 - Ivanhoe B50kb PeterPan 63.5 - 36.5 +35/=57/-8 63.50%

100 Games: http://www.mediafire.com/?b0vakfw5sorodpy

Houdini 3.0 Pro is a real monster. Some of its games are really impressive. A true pleasure for the computer chess lovers. Anyway my favourite Ivanhoe B50kb by PeterPan has scored a very decent 36.5% against him. Almost at Critter 1.6a level!.

Regards from Barcelona.

Tom.

This is some very good work you have done!
I am partial to B50kB as well, and I got these results from running each in 50-game matches ag. 4 common opponents. Unless scaling became a factor for reasons unknown to me, you got quite a few that it would be interesting to see how much over 36.5% they could score: (These were all 1CPU matches, and as you can see I avoid KLO's compiles.)

Code: Select all

Rank Name                       Elo    +    - games score oppo. draws
   1 Ivanhoe 46h x64             41   36   36   200   57%     1   50%
   2 Ivanhoe B46fE.02 x64        28   35   35   200   55%     1   56% 
   3 Ivanhoe B46fE.01 x64        21   35   35   200   54%     1   58% 
   4 Ivanhoe B46e x64            18   37   36   200   53%     1   46%
   5 Ivanhoe B46fC x64           10   35   35   200   52%     1   58%
   6 Ivanhoe B49jAx64p            2   36   35   200   50%     1   55%
   7 Ivanhoe B46a x64             0   36   36   200   50%     1   49%
   8 Ivanhoe B47f0.2 x64         -6   36   36   200   49%     1   49%
   9 Ivanhoe B50kBx64p           -8   36   36   200   49%     1   50%
  10 Ivanhoe B47e x64           -12   36   36   200   48%     1   50% 

Best-

george

[Tomcass]

TESTING Houdini 3 Pro = 60 minutes/game

I7 980 3.33 Ghz.
6 real cores
GUI: Fritz 12
Book: Perfect 2012
Time control: 60 minutes/game
Ponder: Off
No tablebases

Above 50 BNG (Billion Nodes per Game for Houdini).

201211Houdini3Pro 2012

Houdini 3 Pro x64_6 - Critter 1.6 64-bitx6_nob 10.0 - 10.0 +4/=12/-4 50.00%
Houdini 3 Pro x64_6 - Deep Rybka 4.1 SSE42 x64 (x6) 15.5 - 4.5 +12/=7/-1 77.50%
http://www.mediafire.com/?ih09hctkh1mho17

At this time control Critter 1.6 has been a very tough rival for Houdini 3 Pro. It produced some very nice positional games.

Regards,

Tom.

[Tomcass]

Sorry George, I did not realize that you posted this comment. Thank you very much for it.

I don't know why but this old Peterpan compile flies in my older computer with 4 cores. You can see the average nodes per move. It is very difficult to understand for me also. Working in a newer computer with six cores the speed is good, but not so exceptional.

Misteries of computer science.

Best regards from Barcelona!

Tom.