-
Sub-task
-
Resolution: Fixed
-
Normal
-
None
-
None
-
None
-
arches x86-64bit and pure64
sys-devel/gcc:4.9
This is a question rather than a request...
The release notes for gcc 4.9 (link) mention for IA-32/x86-64
- GCC now supports the new Intel microarchitecture named Silvermont through -march=silvermont.
- GCC now supports the new Intel microarchitecture named Broadwell through -march=broadwell.
- Optimizing for other Intel microarchitectures have been renamed to -march=nehalem, westmere, sandybridge, ivybridge, haswell, bonnell.
Comparing the Intel 386 and AMD x86-64 -march options for gcc-4.9.2 (4.9.2 link) with those for gcc-4.8.4 (4.8.4 link) suggests (to me) the following differences in -march values:
| 4.8.x -march | 4.9.x -march |
|---|---|
| corei7 | nehalem |
| * | westmere |
| corei7-avx | sandybridge |
| core-avx-i | ivybridge |
| core-avx2 | haswell |
| atom | bonnell |
* The 4.9.x -march="westmere" appears to be a close match to the 4.8.x -march="corei7-avx" with the exception that it lacks AVX instructions.
Merging sys-devel/gcc-4.9.2-r1 (slotted, with 4.8.4 currently merged) I see that the relevant x86-64bit and pure64 subarch make.defaults continue to have CFLAGS with -march set to the gcc-4.8.x convention.
So my question is that do the make.defaults for these subarches require changing in order for sys-devel/gcc-4.9.x to operate correctly?
