Use -mavx2 -mfma instead of -march with AVX2
commit062a6b81498b61b2bfc4ec7441b844d76aae445b
authorSzilárd Páll <pall.szilard@gmail.com>
Wed, 18 Oct 2017 15:01:54 +0000 (18 17:01 +0200)
committerSzilárd Páll <pall.szilard@gmail.com>
Tue, 7 Nov 2017 20:20:34 +0000 (7 21:20 +0100)
tree0e12214b14036d23ec607958872bd49fb139fc48
parent591f8dd31f4cabcc894f48fc48bded070a58d026
Use -mavx2 -mfma instead of -march with AVX2

This was (likely) only a workaround for some early gcc version that did
not support correct AVX2 code-generation with just the -mavx2 -mfma
flags. However, just as with other SIMD flavors with AVX2 too we should
not request arch-specific tuning just to get the desired SIMD flavor
enabled.

Change-Id: Ib0c6388bebcffbf0719b438451d3943f51fba4a4
cmake/gmxManageSimd.cmake