TalkChess.com

IWB wrote:The current result can be found here:

http://www.inwoba.de

Have fun
Ingo

HOLY CRIMINY!!! HOUDINI 1.5 => 3011 so far!!

http://www.inwoba.de/match.html


Houdini 1.5

Houdini 1.5 - Deep Rybka 4 (2953) 55.5 - 39.5 58.42% Perf=3012
Houdini 1.5 - Stockfish 1.9.1 JA (2897) 59.0 - 36.0 62.11% Perf=2982
Houdini 1.5 - Critter 0.90 SSE42 (2891) 63.0 - 32.0 66.32% Perf=3008
Houdini 1.5 - Naum 4.2 (2821) 76.0 - 19.0 80.00% Perf=3061
Houdini 1.5 - Komodo 1.2 JA (2805) 70.0 - 25.0 73.68% Perf=2983
Houdini 1.5 - Deep Shredder 12 (2800) 73.0 - 22.0 76.84% Perf=3008
Houdini 1.5 - Gull 1.1 (2792) 77.5 - 17.5 81.58% Perf=3050
Houdini 1.5 - Deep Sjeng c't 2010 32b (2773) 74.5 - 20.5 78.42% Perf=2997
Houdini 1.5 - spark-1.0 SSE42 (2753) 80.5 - 14.5 84.74% Perf=3050
Houdini 1.5 - HIARCS 13.1 MP 32b (2732) 82.5 - 12.5 86.84% Perf=3059
Houdini 1.5 - Zappa Mexico II (2713) 82.0 - 13.0 86.32% Perf=3032
Houdini 1.5 - Protector 1.3.2 JA (2701) 82.0 - 13.0 86.32% Perf=3020
Houdini 1.5 - Deep Onno 1-2-70 (2686) 85.5 - 8.5 90.96% Perf=3087
Houdini 1.5 - Hannibal 1.0a (2682) 83.5 - 11.5 87.89% Perf=3026
Houdini 1.5 - Deep Junior 12 (2681) 84.5 - 10.5 88.95% Perf=3043
Houdini 1.5 - Toga II 1.4 beta5c BB (2671) 86.5 - 8.5 91.05% Perf=3074
Houdini 1.5 - Umko 1.1 SSE42 (2632) 87.0 - 8.0 91.58% Perf=3046
Houdini 1.5 - Loop 13.6/2007 (2615) 91.0 - 4.0 95.79% Perf=3157
Houdini 1.5 - Jonny 4.00 32b (2615) 86.5 - 8.5 91.05% Perf=3018

1480.0 - 324.0 82.04% Perf=3011

Long live the Houdini chess engine, the new King of Chess!

Giant congrats to progammer Robert Houdart for a tremendous accomplishment.

notyetagm wrote:

IWB wrote:The current result can be found here:

http://www.inwoba.de

Have fun
Ingo

HOLY CRIMINY!!! HOUDINI 1.5 => 3011 so far!!

http://www.inwoba.de/match.html


Houdini 1.5

Houdini 1.5 - Deep Rybka 4 (2953) 55.5 - 39.5 58.42% Perf=3012
Houdini 1.5 - Stockfish 1.9.1 JA (2897) 59.0 - 36.0 62.11% Perf=2982
Houdini 1.5 - Critter 0.90 SSE42 (2891) 63.0 - 32.0 66.32% Perf=3008
Houdini 1.5 - Naum 4.2 (2821) 76.0 - 19.0 80.00% Perf=3061
Houdini 1.5 - Komodo 1.2 JA (2805) 70.0 - 25.0 73.68% Perf=2983
Houdini 1.5 - Deep Shredder 12 (2800) 73.0 - 22.0 76.84% Perf=3008
Houdini 1.5 - Gull 1.1 (2792) 77.5 - 17.5 81.58% Perf=3050
Houdini 1.5 - Deep Sjeng c't 2010 32b (2773) 74.5 - 20.5 78.42% Perf=2997
Houdini 1.5 - spark-1.0 SSE42 (2753) 80.5 - 14.5 84.74% Perf=3050
Houdini 1.5 - HIARCS 13.1 MP 32b (2732) 82.5 - 12.5 86.84% Perf=3059
Houdini 1.5 - Zappa Mexico II (2713) 82.0 - 13.0 86.32% Perf=3032
Houdini 1.5 - Protector 1.3.2 JA (2701) 82.0 - 13.0 86.32% Perf=3020
Houdini 1.5 - Deep Onno 1-2-70 (2686) 85.5 - 8.5 90.96% Perf=3087
Houdini 1.5 - Hannibal 1.0a (2682) 83.5 - 11.5 87.89% Perf=3026
Houdini 1.5 - Deep Junior 12 (2681) 84.5 - 10.5 88.95% Perf=3043
Houdini 1.5 - Toga II 1.4 beta5c BB (2671) 86.5 - 8.5 91.05% Perf=3074
Houdini 1.5 - Umko 1.1 SSE42 (2632) 87.0 - 8.0 91.58% Perf=3046
Houdini 1.5 - Loop 13.6/2007 (2615) 91.0 - 4.0 95.79% Perf=3157
Houdini 1.5 - Jonny 4.00 32b (2615) 86.5 - 8.5 91.05% Perf=3018

1480.0 - 324.0 82.04% Perf=3011

With less then 75 games to go, it looks like Houdini 1.5 will be rated over 3000 and be 60+ ELO better then version 1.03a.

mwyoung wrote:

With less then 75 games to go, it looks like Houdini 1.5 will be rated over 3000 and be 60+ ELO better then version 1.03a.

WOW!!!

SzG wrote:If I were Vasik I would loan my Rybka code to Robert Houdart. He could surely make a 3200 engine from an original code if he could make a 3100 engine from Robbolito.

Now imagine if Fruit code was given to Robert in 2005. He would have made 3500 elo engine so far .

3011(3010 after 1873 games out of 1900) is probably misleading and I guess that it is translated to 3003(the numbers that you see after part of the games are usually 5-10 elo too high because the calculation is not based on the same method as the final numbers)

I am based on the following from the IPON thread:
"At 1006 games I made a real calculation with bayeselo. The result was 7 Elo lower than the "live rating""

Note that 50 elo advantage relative to Rybka4 and 54 elo improvement relative to previous version is also good enough.

Milos wrote:

SzG wrote:If I were Vasik I would loan my Rybka code to Robert Houdart. He could surely make a 3200 engine from an original code if he could make a 3100 engine from Robbolito.

Now imagine if Fruit code was given to Robert in 2005. He would have made 3500 elo engine so far .

Milos wrote:

SzG wrote:If I were Vasik I would loan my Rybka code to Robert Houdart. He could surely make a 3200 engine from an original code if he could make a 3100 engine from Robbolito.

Now imagine if Fruit code was given to Robert in 2005. He would have made 3500 elo engine so far .

Yes, Houdart would have gone straight from Fruit to Houdini, cutting out the middleman (Rybka).

Le gasp! The thing is incredibly strong! Wheeeeee! Wot fun eh?

notyetagm wrote:

Milos wrote:

SzG wrote:If I were Vasik I would loan my Rybka code to Robert Houdart. He could surely make a 3200 engine from an original code if he could make a 3100 engine from Robbolito.

Now imagine if Fruit code was given to Robert in 2005. He would have made 3500 elo engine so far .

Yes, Houdart would have gone straight from Fruit to Houdini, cutting out the middleman (Rybka).

I think you missed that Milos was taking a swipe at SzG. His "idea" is illogical because Fruit 2.1 was already open-source in 2005.