All builds are Gcc/Mingw compiler except normal Windows 64 bit SSE 4.2 which is Intel compiler.
I included the Intel version for AMD compatibility. GCC SSE4.2 compile did not run on AMD Phenom with last release.
It does now so Intel version may not be needed anymore.
Jim.
OK, that is clear now. We should all ignore the normal Windows 64 bit SSE 4.2 Intel compile. Perfect, now I can move forward.
Needs to be pulled anyway. It did not work right. The non-sse42 64bit ran at 4,884,106 nps at depth of 22. The sse42 64bit ran same at 4,760,583 nps, which means it is slower. Shouldn't be. Whereas the GCC sse42 depthed out at 5,041,968 nps. Which is the fastest- but not by enough to really matter either way. Now the 32bit showed 3,491,284 nps at same depth. You can lose a good bit if you have to use 32 bit- personally- what I am looking at says it doesn't much matter whether you use GCC sse42 x 64, or just the plain x64. You ain't gonna see any difference in actual play strengthwise either way.
Yes, with the GCC SSE42 version being just 3% faster than the generic x64 version, you will never see any Elo gain because if it exists, it will be lost in the error margins. But if you are doing analysis of some sort, then by all means take the 3%.
Hi Jim,
I have some problems to understand the discussion. Can you please cofirm, which 64bit compile I can use on my core I7 pc:
stockfish-23-64-sse42-ja.exe 802.304 15.09.2012 15:05 -a--
stockfish-23-64-ja-sse42-i7.exe 815.616 15.09.2012 14:44 -a--
stockfish-23-64-ja.exe 821.760 15.09.2012 14:14 -a--
stockfish-23-w32-ja.exe 800.256 15.09.2012 10:01 -a--
What has changed? We noticed that there are significant differences in 2.2.2 and this version with respect to playing characteristic.
For example, on a 1 ply search version 23 is about 30 ELO stronger when playing head to head (only 8000 games) and is over twice as fast.
That implies some evaluation improvements and either quies search performance improvements or perhaps simply startup costs have been reduced. One version of Komodo was very slow on sub 5 ply searches because we spend a lot of time obsessing over the ply 1 move list.
However 2.3 seems to retains a big speed advantage although not the 2 to 1 I see in 1 ply searches. It searches at least 1/4 ply more.
Can you tell us what the most impactful changes are?
mcostalba wrote:Thanks Jim, very kind and helpful as usual.
Just a couple of quick notes.
1) For people compiling themself the Stockfish bench signature is: 5416292 It means that if you run ./stockfish bench then it should show 5416292 nodes searched. This means that the compile is correct.
2) For people running / using under Android / iPhone with more than one thread, I'd suggest to set "Use Sleeping Threads" option. This allows to save power and battery and strength is practically the same.
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.
Werner wrote:Hi Jim,
I have some problems to understand the discussion. Can you please cofirm, which 64bit compile I can use on my core I7 pc:
stockfish-23-64-sse42-ja.exe 802.304 15.09.2012 15:05 -a--
stockfish-23-64-ja-sse42-i7.exe 815.616 15.09.2012 14:44 -a--
stockfish-23-64-ja.exe 821.760 15.09.2012 14:14 -a--
stockfish-23-w32-ja.exe 800.256 15.09.2012 10:01 -a--
Which is not ok?
Hi Werner,
There were two SSE 4.2 Windows compiles. GCC version for i7 is ok, delete the other one (which was Intel) or just download again from my homepage. I have removed the bad one.
Werner wrote:Hi Jim,
I have some problems to understand the discussion. Can you please cofirm, which 64bit compile I can use on my core I7 pc: stockfish-23-64-sse42-ja.exe 802.304 15.09.2012 15:05 -a--
stockfish-23-64-ja-sse42-i7.exe 815.616 15.09.2012 14:44 -a--
stockfish-23-64-ja.exe 821.760 15.09.2012 14:14 -a--
stockfish-23-w32-ja.exe 800.256 15.09.2012 10:01 -a--
Which is not ok?
Hi Werner,
There were two SSE 4.2 Windows compiles. GCC version for i7 is ok, delete the other one (which was Intel) or just download again from my homepage. I have removed the bad one.
Jim.
Thanks Jim,
as I saw you deleted the red marked one now. Seems to be a bit slower...