GCC 4.6 Compiler Performance With AVX On Sandy Bridge
Written by Michael Larabel in Processors on 7 February 2011. Page 4 of 6. 16 Comments

POV-Ray, another interesting ray-tracing application, was a bit more interesting even though it is currently single-threaded. The performance improved greatly when switching from GCC 4.3 to GCC 4.5, but it improved even more when using GCC 4.6 with the core2 flags, but it didn't get any better with the SSE4 (Core i7) or AVX (Core i7 Sandy Bridge) support.

The Himeno test results were largely uninteresting and stable except for when enabling the AVX floating-point math support, which for this test negatively affected its performance.

With MAFFT, enabling the AVX floating-point math support in the GNU Compiler Collection led to a slight decrease in time needed for this multiple sequence alignment search.

Related Articles
Featured Articles
Trending Linux News