Stockfish 2.2.2 JA update available

[noctiferus]

Sts summary results:

Exact moves: 1118 / 1300 Grade : S
Score: 11780 / 1300 grade S
More details on
http://www.open-chess.org/viewtopic.php?f=7&t=1818

[pohl4711]

The new Version 2.2.2 produces 35% time losses (!!!), like the older 2.2. Only the 2.2.1 works correct.
My conditions: Old Intel Quad (no SSE support), Vista 64bit, LittleBlitzerGUI, 1 core per engine, 60 seconds + 500ms Fischerbonus.

Stefan

[mcostalba]

The new Version 2.2.2 produces 35% time losses (!!!), like the older 2.2. Only the 2.2.1 works correct.
My conditions: Old Intel Quad (no SSE support), Vista 64bit, LittleBlitzerGUI, 1 core per engine, 60 seconds + 500ms Fischerbonus.

Thanks Stefan for your report. Could you please verify if also the universal version 2.2.2 JA (not SSE) produces time losses ?

[pohl4711]

The new Version 2.2.2 produces 35% time losses (!!!), like the older 2.2. Only the 2.2.1 works correct.
My conditions: Old Intel Quad (no SSE support), Vista 64bit, LittleBlitzerGUI, 1 core per engine, 60 seconds + 500ms Fischerbonus.

Thanks Stefan for your report. Could you please verify if also the universal version 2.2.2 JA (not SSE) produces time losses ?

At the moment the RobboLito 0.10 blitz-testrun for my NEBB-Rankinglists is in progress. It will end saturday-morning. Then I can test another Stockfish version for time-loss-problems, but what is the "universal"-Version and where can I download it?

Stefan

[Jim Ablett]

The new Version 2.2.2 produces 35% time losses (!!!), like the older 2.2. Only the 2.2.1 works correct.
My conditions: Old Intel Quad (no SSE support), Vista 64bit, LittleBlitzerGUI, 1 core per engine, 60 seconds + 500ms Fischerbonus.

Thanks Stefan for your report. Could you please verify if also the universal version 2.2.2 JA (not SSE) produces time losses ?

At the moment the RobboLito 0.10 blitz-testrun for my NEBB-Rankinglists is in progress. It will end saturday-morning. Then I can test another Stockfish version for time-loss-problems, but what is the "universal"-Version and where can I download it?

Stefan

Hi Stefan,

If you download Stockfish from my site it also contains in addition, an Intel folder (a well as a GCC folder). The official GCC compiles (Posix threads) are the ones which I think have the time issues. The Intel compiles (native Windows threads) include the time fix from 2.2.1.

The non-sse 4.2 builds are the universal ones. (which I think you are using already)

Jim.

[Jouni]

Too many versions, I am complete lost now! Probably 2.2.1 is still the strongest

Jouni

[pohl4711]

The Intel compiles (native Windows threads) include the time fix from 2.2.1.

OK, found it. Will try it this weekend and report it here.

Stefan

[Werner]

Hi Marco,
I had here a match with Stockfish 2.2.2 x64 1CPU SSE version. In 300 games I have 45 losses in time after move 40 (80, 120)
emergency move time is 70 (standard) - would it help to go to 100 ?

best wishes
Werner

[Werner]

Hi Marco,
I had here a match with Stockfish 2.2.2 x64 1CPU SSE version. In 300 games I have 45 losses in time after move 40 (80, 120)
emergency move time is 70 (standard) - would it help to go to 100 ?
best wishes
Werner

Correctíon: I have used the first downloads which contains only the gcc compiles (the engine has 1.000.448).
I will now try the Intel compiles and hope they have the same speed.

[Werner]

Problem solved!
all 45 games repeated without losses in time!
And I have had much better results with Stockfish 2.2.2 here - so next week more games with this version.
Werner