[FX]: theNonsuch: 1.0 Final [Aviary / OS X] (G5 optimized)

krmathis · Post by **krmathis** » February 9th, 2005, 12:39 am

Quevar, that sounds interesting.
I`ll compile a "-mcpu=7450 -fast" build one of these days, then compare the speed to my regular optimized build!

Quevar · Post by **Quevar** » February 9th, 2005, 10:18 am

I can't really say if it is faster cause it is blazingly fast the original way your wrote it. My experinece with compiling programs with the -fast option results in about 2-3 times faster than only "-O3 -faltivec", but these tend to be highly numerical number crunching programs. Since Firefox isn't really a number crunching program, I don't know if there is going to be huge difference. I'll try the "-mcpu=7450 -fast" when I get home on my G4 - maybe it will be 'slow' enough that I'll be able to tell a difference.

This is my first time compiling Firefox, so I'm far from an expert on this type of program. I just thought it would be interesting to try it out.

krmathis · Post by **krmathis** » February 9th, 2005, 11:24 am

@Quevar. If you really want to speed test different compiler profiles with a number crunching application, give the LZMA SDK at test! The standalone lzma compressor have a Benchmark feature.

Download the 4.06 version, unpack and cd to the following directory: lzma406/SRC/7zip/Compress/LZMA_Alone/
Add you own compiler option to the makefile, then run "make" to compile ("make clean" to clean up afterwards).