Stockfish 2.2.1 update available

[IWB]

Not good
Compile is dropping out of SSE4.2 (with popcount support) and running in a lower mode SSE3 or SSE4 (without popcount) that's why no speed increase and no benefit over universal compile.

Ok one more try. Same link as before >

http://dl.dropbox.com/u/5047625/stockfi ... cks-ja.zip

Jim.

This version is showing exactly the same like before! THe original SSE42 (with the new locks) is showing exacrtly the same on an i7 with W7-64. At the moment it looks like th ewhole POP on Phenom2 AND i7 is brocken somehow ...

BYe
Ingo

[Engin]

i do'nt know if is broke but its run and display it with SSE4.2 without crashes

[Jim Ablett]

Hmm, the origiinal SSE42 on W7 and i7 is showing this:

Stockfish 2.2.1 JA 64bit SSE4.2 by Tord Romstad, Marco Costalba and Joona Kiiski
uci
id name Stockfish 2.2.1 JA 64bit SSE4.2
id author Tord Romstad, Marco Costalba and Joona Kiiski

option name Use Search Log type check default false
option name Search Log Filename type string default SearchLog.txt
option name Book File type string default book.bin
option name Best Book Move type check default false
option name Mobility (Middle Game) type spin default 100 min 0 max 200
option name Mobility (Endgame) type spin default 100 min 0 max 200
option name Passed Pawns (Middle Game) type spin default 100 min 0 max 200
option name Passed Pawns (Endgame) type spin default 100 min 0 max 200
option name Space type spin default 100 min 0 max 200
option name Aggressiveness type spin default 100 min 0 max 200
option name Cowardice type spin default 100 min 0 max 200
option name Min Split Depth type spin default 4 min 4 max 7
option name Max Threads per Split Point type spin default 5 min 4 max 8
option name Threads type spin default 4 min 1 max 32
option name Use Sleeping Threads type check default false
option name Hash type spin default 32 min 4 max 8192
option name Clear Hash type button
option name Ponder type check default true
option name OwnBook type check default true
option name MultiPV type spin default 1 min 1 max 500
option name Skill Level type spin default 20 min 0 max 20
option name Emergency Move Horizon type spin default 40 min 0 max 50
option name Emergency Base Time type spin default 200 min 0 max 30000
option name Emergency Move Time type spin default 70 min 0 max 5000
option name Minimum Thinking Time type spin default 20 min 0 max 5000
option name UCI_Chess960 type check default false
uciok

So no popcount as well ... now I am confused. Does the SSE work at all?

Bye
Ingo

Sorry I just checked the src. Looks like 'popcount detected' message has been taken out in 2.2.1. Now If it show's SSE4.2 in startup string then it is using popcount. Marco has run The SSE4.2 i7 build and reported a 4% speed up over the universal.

Jim.

[IWB]

OK, I will run all 3 compiles to compare:

1. x64
2. x64 SSE42
3. x64 SSE42 oldlocks

If everything is allright I should see a difference between 1 to 2 and 3 but not between 2 and 3 (or very little).

Wait 10 min.

Bye
Ingo

[tomgdrums]

For what it is worth I could not get the new SSE update to work in Windows 7 with the Shredder GUI. The normal 64 bit version worked fine. The Shredder GUI just kept saying the SSE version was not a UCI engine.

[Jim Ablett]

For what it is worth I could not get the new SSE update to work in Windows 7 with the Shredder GUI. The normal 64 bit version worked fine. The Shredder GUI just kept saying the SSE version was not a UCI engine.

Maybe I should have just stuck with the universal 64 bit compile. This is all getting a bit messy just for extra 4% speedup.

Jim.

[IWB]

For what it is worth I could not get the new SSE update to work in Windows 7 with the Shredder GUI. The normal 64 bit version worked fine. The Shredder GUI just kept saying the SSE version was not a UCI engine.

Maybe I should have just stuck with the universal 64 bit compile. This is all getting a bit messy just for extra 4% speedup.

Jim.

OK, I'll try to sumarize:

All my test where done with th eCLassic GUI, the latest versions run fine!

On W7-x64 and an i7-920 CPU the x64, the x64-SSE version AND the x64-SSE-Oldlocks run fine!

1. x64:
Engine: Stockfish 2.2.1 JA 64bit (256 MB)
by Tord Romstad, Marco Costalba and Joona

27/40 5:19 +0.28++ 1.Nf3 Nf6 2.d4 e6 3.e3 d5 4.c4 Be7
5.cxd5 exd5 6.Bd3 O-O 7.O-O c5 (498.386.004) 1557


2. x64-SSE:
Engine: Stockfish 2.2.1 JA 64bit SSE4.2 (256 MB)
by Tord Romstad, Marco Costalba and Joona

27/40 5:02 +0.28++ 1.Nf3 Nf6 2.d4 e6 3.e3 d5 4.c4 Be7
5.cxd5 exd5 6.Bd3 O-O 7.O-O c5 (498.386.004) 1647

3. x64-SSE-oldlocks
Engine: Stockfish 2.2.1 JA 64bit SSE4.2 oldlock (256 MB)
by Tord Romstad, Marco Costalba and Joona

27/40 5:02 +0.28++ 1.Nf3 Nf6 2.d4 e6 3.e3 d5 4.c4 Be7
5.cxd5 exd5 6.Bd3 O-O 7.O-O c5 (498.386.004) 1645

Conclusion:

SSE version is about 5% faster than the non SSE version. The Vista/W7 version has the same speed as the XP-oldlock version!

The only problem you had was to believe that the 'new' Visto/W7 is any faster than the 'oldlocks' thing.

If I where you I would replace the Vista/W7 with the oldlocks and everything is fine!

Thanks for your efforts, I appreciate that and will test now th e"oldlocks" SSE version!

Bye
Ingo

[Engin]

i just tested all versions 32/64 bit and SSE4 with shredder classic 4 c't and they running very well on me.

[Jim Ablett]

Conclusion:

SSE version is about 5% faster than the non SSE version. The Vista/W7 version has the same speed as the XP-oldlock version!

The only problem you had was to believe that the 'new' Visto/W7 is any faster than the 'oldlocks' thing.

If I where you I would replace the Vista/W7 with the oldlocks and everything is fine!

thanks for testing Ingo. I will replace with old locks version.

Is the oldlock version you tested the last one I compiled here (I've recompiled it 3 times now) >

http://dl.dropbox.com/u/5047625/stockfi ... cks-ja.zip

Jim.

[IWB]

thanks for testing Ingo. I will replace with old locks version.

Is the oldlock version you tested the last one I compiled here (I've recompiled it 3 times now) >

http://dl.dropbox.com/u/5047625/stockfi ... cks-ja.zip

Jim.

YES, I checked and tested again! This version is fine on W764 AND XP64 and it is working as SSE as it is faster than the universal x64!

BYe
Ingo