Critter 1.0 X64 SSE4.2 test

Critter 1.0 X64 SSE4.2 vs Houdini 1.5a X64. Testing done on 980X CPU HT on. 6 vs 6 threads.
TC = 1' +1", ponder on, 512 hash, rybka 4 book, Reverse colors.

Results.

Critter 1.0 X64 SSE4.2 3 wins, 14 losses, 13 draws. -133 elo below Houdini.

More than likely this is a statistical fluke. The engine scored worse than expected. Its probably around 90-100 elo behind Houdini.

Very interesting play style. The engines seems to steer towards imblanced endgames and should not be underestimated for playing into those types of postions. I hope the author is able to make great strides to improves this engine further. It might one day emerge as the top engine if the auther keeps making dramatic improvements. Like what he has done over the last few engine releases.

[You must be registered and logged in to see this link.] wrote:

Critter 1.0 X64 SSE4.2 vs Houdini 1.5a X64. Testing done on 980X CPU HT on. 6 vs 6 threads.
TC = 1' +1", ponder on, 512 hash, rybka 4 book, Reverse colors.

Results.

Critter 1.0 X64 SSE4.2 3 wins, 14 losses, 13 draws. -133 elo below Houdini.

More than likely this is a statistical fluke. The engine scored worse than expected. Its probably around 90-100 elo behind Houdini.

Very interesting play style. The engines seems to steer towards imblanced endgames and should not be underestimated for playing into those types of postions. I hope the author is able to make great strides to improves this engine further. It might one day emerge as the top engine if the auther keeps making dramatic improvements. Like what he has done over the last few engine releases.

like to see longer time control test without that awfull book , and ensureing niether has large pages or bases

say 3+1

I might be able to do one tonight at 3' +1" Time Control but, what book do you think i should use? I also forgot to mention that I was using the Gaviota Table bases since both engines were able to use the same TB's. Also critters eval of postions is not really all that acurate. It kinda reminds me of how Stockfish evals are kinda flakey.

[You must be registered and logged in to see this link.] wrote:

I might be able to do one tonight at 3' +1" Time Control but, what book do you think i should use? I also forgot to mention that I was using the Gaviota Table bases since both engines were able to use the same TB's. Also critters eval of postions is not really all that acurate. It kinda reminds me of how Stockfish evals are kinda flakey.

"B" finds no books . no bases , no large pages best format to fully test an engine

hmmmmm. But with no book there is lack of variations. Engines will only vary openings so much and its quite limited in scope. I don't really see how this would help. Since you would get way to many repeated openings following the same lines of play. Computers are vary narrow minded people :)

[You must be registered and logged in to see this link.] wrote:

hmmmmm. But with no book there is lack of variations. Engines will only vary openings so much and its quite limited in scope. I don't really see how this would help. Since you would get way to many repeated openings following the same lines of play. Computers are vary narrow minded people :)

"B" says she finds far more variation of unsual openings with no books , especialy D00 A00 etc

shes used no books no base for sometime now and says its the only real way of testing the engines capabilities , she preferes longer tc but ive read vas empowered his clones via similar testings only ultra vast ones , very many thousands of them
i only know that in book testing different tc give completely different results and lower tc poroduce far more blunders

test as you feel best , mine was only a suggestion and we are appreciative of all tests

I'll be glad to try a no book to see what happens.

Computer: i7 975 4 real cores + 4 Hyper Thread.
Fritz Benchmark: Speed 26,73 KNS 12.828
Gui: Fritz 12
Time Control: 3' + 1
Book: H.S. Masterbook 3.1
Ponder: Off.
No Robbo, Triple or Nalimov.
No crashes.


201103.Critter1.0-1 2011

Critter 1.0 64-bit SSE4(x4) - Houdini 1.5a x64(x4) 11.5 - 18.5 +4/=15/-11 38.33%
Critter 1.0 64-bit SSE4(x4) - Naum 4.2 64(x8) 18.0 - 12.0 +10/=16/-4 60.00%
Critter 1.0 64-bit SSE4(x4) - Ivanhoe B50kBx64p(x4) 16.5 - 13.5 +7/=19/-4 55.00%
Critter 1.0 64-bit SSE4(x4) - Deep Rybka 4 x64 (x4) 13.0 - 17.0 +4/=18/-8 43.33%