| Subject: Re: Improve your BOINC/Seti performance |
| From: Martin 53N 1W |
| Date: 04/08/2004, 18:53 |
Devon Ferns wrote:
If you have access to VS.NET 7, compiling specifically for the P4 if you
have one makes a pretty big difference in the BOINC CPU benchmarks.
P4 system is 1.4GHz.
For example, compiling it just for the P4 and above resulted in my
benchmarks going from 852 to 954 Whetstone and from 1688 to 1941 Dhrystone.
That's a pretty good improvement for just recompiling the program. I also
[...]
Good note.
For those using gcc on linux with an Athlon Palomino core CPU, try
recompiling with:
# Set the default optimisation flags for gcc
# Full optimisations for the AthlonXP on gcc 3.2.2
export MACHTYPE="i686-mandrake-linux-gnu"
export HOSTTYPE="i686"
export CHOST="i686-pc-linux-gnu"
export CFLAGS="-march=athlon-4 -O3 -funroll-loops -mfpmath=sse
-minline-all-stringops -falign-functions=4 -frerun-cse-after-loop
-frerun-loop-opt -m3dnow -malign-double -mfancy-
math-387 -mfp-ret-in-387 -pipe -fschedule-insns2 -fexpensive-optimizations"
export CXXFLAGS="$CFLAGS"
I keep away from the 'unsafe math' flags. Saving a few more cycles is
not worth losing the floating point math results checks.
Anyone with their best settings for any other systems?
Regards,
Martin