gcc and avx

gcc and avx

Can someone post an example of how to write an accumlate (for eight 32bit regs) using the vector instructions in AVX?

It would be best if the code could be written with gcc (extensions are ok).

3 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

You would require the ix86/avx branch of gcc 4.4 http://www.gnu.org/software/gcc/svn.html AVX support presumably is mainly floating point; ideally, a float sum reduction in C should work with -ffast-math (doubt if it is so yet).

GCC-v4.4 & Binutils(v-2.18.50.0.9) has AVX support, but remember it's just 128-bit extension currently by GCC.

Leave a Comment

Please sign in to add a comment. Not a member? Join today