From a software perspective, how do you guarantee that an engine's algorithm doesn't appropriate cpu/ram/cache resources in an unfair manner while running alongside the other engine?
Rybka vs Houdini 100-game match results
Sort:
Shivsky wrote:
From a software perspective, how do you guarantee that an engine's algorithm doesn't appropriate cpu/ram/cache resources in an unfair manner while running alongside the other engine?
well, i don't know how to guarantee it. i did monitor the engines in windows performance monitor--which watches cpu/ram/disk/etc--before running the tournament. they seemed to be switching on and off 100% when it was or wasn't their turn. try it.
Houdini is faster so it's better for quick analysis
Vladan88 wrote:
Houdini is faster so it's better for quick analysis
After running this match and watching them grind away at similar positions in games 51-100, I can say I fully agree with the statement that it is faster. better, though? depends on the position.
What would you say is the most interesting game (or games) of all these? Has anyone done an analysis of them?
Were there any novelties by the computers?
bfinn wrote:
What would you say is the most interesting game (or games) of all these? Has anyone done an analysis of them?
most interesting? can't really say. depends on people's preferences, really. there were interesting *moments*, though. some really surprising moves. i can't remember any specific ones right now. i should make a web page for each win, so people can see the games. hmmm... i would love for people to analyze them, do an eco breakdown, etc.
Vyomo wrote:
Were there any novelties by the computers?
yes. several. i am sorry i can't remember them. i think i will create a web page for each win, so people can watch the games w/o having to cut and paste the PGNs and view them themselves.
ivoryknight71 wrote:
The Deep Rybka 4.1 SSE42 x64 vs Houdini 1.5a x64 100 - game match with long time controls is over.
[...]
See the games here: http://www.westportchessclub.org/computer-chess/rybka-vs-houdini-100-games-match.htm
The link does not work any more. Any idea where the PGN of these games could be found?
im not able to access the games anymore =(.
The Deep Rybka 4.1 SSE42 x64 vs Houdini 1.5a x64 100 - game match with long time controls is over.
----------
Final Score: Rybka 51 - Houdini 49
Format: 2 hours for first 40 moves, 2 hours for next 40 moves, 90 minutes + 30 seconds/move for the rest of the game.
Hardware: AMD Phenom(tm) 9750 Quad-Core Processor with 8 GBs Memory
Operating System: Windows 7 Professional SP1 (Build 7600) 64-bit
Tablebases: Gaviota & Nalimov 3-4-5 piece EGTBs on an SSD
Opening Book: Generic books up to move 10
Hash Size: 2,048 MBs
Ponder: Off
----------
For the first 50 games, the engines chose the openings at random (up to move 10), but they did NOT have to play both White and Black for each opening. For the last 50 games, I chose the openings (up to move 10), and they DID have to play both White and Black for each opening. The score for the first 50 games was a tie. The score for the last 50 games was +2 for Rybka.
See the games here: http://www.westportchessclub.org/computer-chess/rybka-vs-houdini-100-games-match.htm