Komodo64 4 SSE vs Stockfish 2.2.2 JA SSE42 results

[ivoryknight]

Also at the new time control:
http://atomicc-testing.blogspot.com/201 ... 22-ja.html

[ivoryknight]

I forgot Stockfish caused an exception in game 2. I am re-running it now. :-/

[ivoryknight]

Round 2 was a draw. The results have been corrected.

[lkaufman]

I just want to say that I think your test conditions are very good, and your ratings should be a fairly good reflection (with enough games) of the relative strength of the engines at a time control that is reasonably similar to the way most engine owners use their engines. In my view increment testing is the way to go, nobody plays human tournaments at time limits like 40/4 minutes repeating for example. The forty move control primarily serves to allow humans playing long games to get a break to use the restroom and/or get something to drink; it has no place in rapid play or in engine vs engine play. I don't mean by this that lists like CCRL and CEGT are less valid, just that they are wasting resources playing out long dead-drawn endgames at the same slow level as the middlegame. Increment play gives the best balance between time spent in the middlegame and the endgame, and the increment should be at least ten seconds since most people will give their engines at least that much time per move when reviewing a game or analyzing. In sum, I think your 10' + 10" time control is pretty optimal. Good luck with your testing, and passing Houdini on your list looks like a difficult but doable goal for the next Komodo.

[ATOMICC]

...Good luck with your testing, and passing Houdini on your list looks like a difficult but doable goal for the next Komodo.

Thank you for your input, Mr. Kaufman. I wish you the best in your goal with your incredible engine. I eagerly await the MP version. I will put all 6 CPUs to use for the CCRL blitz (and my own testing for fun) then.