Typing bench or analyze and d4 on console mode
causes terminating Crafty.
Best,
PS: Jims compile runs fine, but with much lower performance.
Yes Harun. What I meant before was that the Ming32 build by JA worked using smpnice=1. Cpu usage drops to zero on opponents turn on smp-mode.
My Intel compile is just a test build, though it works fine like the last fix in WB and Arena with smpnice=1. And you could only use smpnice=0 in CB and SC interface. The only difference I saw was that it will not crash when you stop and resume games, not like before.
The Ming32 build by Master Jim is the best so far in compatibility with "ALL" GUIs.
I havent tested Jims latest build and your latest build
in blitz games. So dont know about their performance
in tournaments. Your build is around 50% faster than
gcc build. <knps figures>
On the other hand, I have carried out few test games
with different crafty compiles. All 32 bit, 2 cpu.
Standings so far:
1. Jims first compile,
2. Peters compile,
3. Your compile,
4. Jims second compile
Is not interesting?
There are 78 points between Jims first and second compiles.
With more games, this gap will be smaller, I would think.
I havent tested Jims latest build and your latest build
in blitz games. So dont know about their performance
in tournaments. Your build is around 50% faster than
gcc build. <knps figures>
On the other hand, I have carried out few test games
with different crafty compiles. All 32 bit, 2 cpu.
Standings so far:
1. Jims first compile,
2. Peters compile,
3. Your compile,
4. Jims second compile
Is not interesting?
There are 78 points between Jims first and second compiles.
With more games, this gap will be smaller, I would think.
Best,
Yes, there's no difference testing games with engine compiles built with identical codes. Whichever is faster is more likely having the advantage. Another factor is OS and hardware compatibilities.
My last upload was not built for speed, used all codes (with no editing) and the safest compiler switches I could grasp, just to try if it will work like the MingW compile of Jim. Jim already mentioned before that ICC optimization switches has a negative effect on the smpnice code. The ming32 fix was the best option for this type of problem (at the moment). I'm sure it will be fixed as the discussion continues. Thanks pal.
