thanks to you and of cource to Dan fur this faster Bruja 1.9
i am interested if you have also speedtests for your compiles Resp 0.19 and Scid 3.61
in my tournaments -Div E and Senior - Resp 0.19 did very better (+130), but with Scid 3.61 (default-settings) (+50) i can see not so much increase of strenght as with Resp 0.19
with UseIDD=false, it was also not better
may be there are to less games
so i would like to see what the speedtests say
thanks to you and of cource to Dan fur this faster Bruja 1.9
i am interested if you have also speedtests for your compiles Resp 0.19 and Scid 3.61
in my tournaments -Div E and Senior - Resp 0.19 did very better (+130), but with Scid 3.61 (default-settings) (+50) i can see not so much increase of strenght as with Resp 0.19
with UseIDD=false, it was also not better
may be there are to less games
so i would like to see what the speedtests say
thanks
Hi Gerhard,
Here are the speed comparisons running 'perft 6':
RESP 0.19
Resp 0.19 x64 JA
-----------------------
NODES = 119060324
TIME USED = 13.609
NODES/SEC = 8.74865e+006
Resp 0.19 win32 JA
-----------------------
NODES = 119060324
TIME USED = 20.516
NODES/SEC = 5.80329e+006
Resp 0.19 win32 Original
------------------------
NODES = 119060324
TIME USED = 22.984
NODES/SEC = 5.18014e+006
SCIDLET 3.61
Scidlet 3.61 x64 JA
---------------------
Moves: 119060324 Time: 17.23 s
Scidlet 3.61 win32 JA
-------------------------------
Moves: 119060324 Time: 24.04 s
Scidlet 3.61 win32 Original DC
------------------------------
Moves: 119060324 Time: 29.03 s
With Resp, I included a new book built from Frank Quisinsky's 'gm2600.pgn'. This may also have added some extra elo over the original book, that's if you used it or course!.
I'd be pleased to do some new compiles, but first step is to get hold of the source which is private/closed. I'll contact Miguel. If I can get it, I'll do it.
Gaviota 0.22 performed much better than Gaviota 0.33
in my recent tests: approx. min 200 points.
Could you perhaps take a look at Gaviota's code and
produce new compiles of 0.22 and 0.33?
Gaviota 0.33 is a very good engine. Only one drawback:
After some good 30 moves, 0.33 makes illegal move
and forfeits the game.
Well, a fixed new compile of Gaviota 0.33 would be great.
Best,
This is weird. Gaviota 0.33 is not super strong, but is better than 0.22. Also, it has been around for at least four years and it has been always stable. Anyway, I am interested to fix whatever problem Gaviota has. Are you using winboard? any interface to use it as UCI?
If you detect any problem that can be reproduced, please sent me the logs or anything that can help me detect it. If someone wants to do a compile, I think I can provide the source, but I do not think that is the problem.