GCC 4.6 Compiler Performance With AVX On Sandy Bridge
Written by Michael Larabel in Processors on 7 February 2011 at 09:40 AM EST. Page 3 of 6. 16 Comments.
Regardless of compiler version or flags, the blowfish performance in John The Ripper was virtually unchanged.
The BYTE performance was also largely uninteresting for this compiler / AVX focus.
C-Ray is one of our favorite multi-threaded CPU tests, but this time around its results weren't too interesting besides a performance regression from GCC 4.3.5 to GCC 4.4.5 and its ray-tracing times being the same even when enabling the Core 2 / Core i7 / Sandy Bridge optimizations. It was not until enabling the AVX floating-point math support that this ray-tracing application really pulled ahead to deliver a 23% speed-up.