@NiLuJe, towbob: Looks good, with two notes:
Quote:
|
Originally Posted by gcc-4.6
If the selected floating-point hardware includes the NEON extension (e.g. -mfpu=`neon'), note that floating-point operations will not be used by GCC's auto-vectorization pass unless -funsafe-math-optimizations is also specified. This is because NEON hardware does not fully implement the IEEE 754 standard for floating-point arithmetic (in particular denormal values are treated as zero), so the use of NEON instructions may lead to a loss of precision.
|
Debian and Unbuntu have decided to move forward to newer minimum arm cores and using the half-wide floating point registers option:
-mfp=vfpv3-d16
Both notes above only apply to building code for the K4&K5 (armv7 core).
The other machines, which are actually vfp1.1, are matched by the gcc option 'vfp' (which in fact generates code for vfp1.1, vfp1 code never got into gcc before the change in the cores).