Page 1 of 1

[STS 1-13] Critter 0.90

Posted: Fri Dec 03, 2010 4:23 pm
by Swaminathan
Results of STS 1-13 for
Critter 0.90:
Single CPU, Q6600, 32 bits, Arena 2.5 beta

Code: Select all

STS 01: 94
STS 02: 80
STS 03: 84
STS 04: 89
STS 05: 86
STS 06: 84
STS 07: 83
STS 08: 74
STS 09: 79
STS 10: 85
STS 11: 84
STS 12: 83
STS 13: 84
Total:

Code: Select all

1089/1300

I had sent the logfile to Gailhac to see if STS Stat could be made to work with scores from STS 13 taken into account.


Critter 0.80

Image

In comparison with Critter 0.70

Image

http://sites.google.com/site/strategict ... st-results

1000 Positions
10 seconds per position
Hardware: Q6600, 32 bits, 2 GB RAM, 2.4 GHZ. Arena 2.01 GUI.

Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 4:01 am
by BB+
Stockfish 1.9 only had 975/1200 on the first 12 tests, while Critter 0.90 got 1005.

Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 4:35 am
by kingliveson
Take this with a grain of salt as it's a small sample. Both engines play the same positions as white/black:


Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 4:59 am
by kingliveson
Critter vs Houdini:


Critter vs IvanHoe:

Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 10:23 am
by Swaminathan
BB+ wrote:Stockfish 1.9 only had 975/1200 on the first 12 tests, while Critter 0.90 got 1005.
Frank Quisinsky recommended me to test Stockfish 1.91 (which is reportedly much much better than 1.9)
I hadn't gotten around to testing it. I had only tested 1.9 which is slightly weaker.

From what I've seen

Stockfish 1.91 > Critter 0.9 > Stockfish 1.9

Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 10:27 am
by Swaminathan
Another reason might be that authors of Stockfish don't often tune their engine and optimise it with STS, they just make changes and do few tests.

Richard Vida certainly does both. He constantly reviews performance in both STS and actual tests.

This is perhaps why Critter may be optimised to do better. But it's very small difference though.

Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 10:49 pm
by ernest
Swaminathan wrote:Frank Quisinsky recommended me to test Stockfish 1.91 (which is reportedly much much better than 1.9)
This is stupid : in infinite analysis 1.9 and 1.91 give exactly the same PVs
(which is also what the Stockfish authors said, when giving out the 1.91)

In 64-bit, the 1.91 is even 2-3% faster than 1.9

Re: [STS 1-13] Critter 0.90

Posted: Sat Dec 04, 2010 11:18 pm
by kingliveson
Knew something was missing -- here it is vs Rybka:

Re: [STS 1-13] Critter 0.90

Posted: Sun Dec 05, 2010 3:11 am
by Swaminathan
ernest wrote:
Swaminathan wrote:Frank Quisinsky recommended me to test Stockfish 1.91 (which is reportedly much much better than 1.9)
This is stupid : in infinite analysis 1.9 and 1.91 give exactly the same PVs
(which is also what the Stockfish authors said, when giving out the 1.91)

In 64-bit, the 1.91 is even 2-3% faster than 1.9
In 32 bit, it's even faster,

http://talkchess.com/forum/viewtopic.ph ... 32&t=36565