| 1 |
maxim wexler writes: |
| 2 |
|
| 3 |
> --- On Thu, 5/21/09, Alex Schuster <wonko@×××××××××.org> wrote: |
| 4 |
|
| 5 |
> > But it suggests using -march=k8 - isn't that a 64-bit-only |
| 6 |
> > thing? I'm not |
| 7 |
> |
| 8 |
> For an 900A w/intel Atom? |
| 9 |
|
| 10 |
No, I have an AMD Athlon(tm) Dual Core Processor 4850e. |
| 11 |
|
| 12 |
> This is what I get: |
| 13 |
> |
| 14 |
> Warning: Your compiler supports the -march=native option which you may |
| 15 |
> prefer Warning: Newer versions of GCC better support your CPU with |
| 16 |
> -march=atom -march=core2 -mtune=pentium -mfpmath=sse. |
| 17 |
> |
| 18 |
> Now I'm confused. It says *my* compiler supports -march=native. Then it |
| 19 |
> says "Newer versions". Isn't v4.3.2 new? It was from a new pkg about a |
| 20 |
> month ago. Is it giving me a choice here? Can I really declare two -march |
| 21 |
> variables? What about mtune and mfpath, are they meant to be "instead of" |
| 22 |
> or "in addition to"? |
| 23 |
|
| 24 |
I'm also confused. Multiple -march options make no sense I believe, as does |
| 25 |
specifying both -march and -mtune, as -mtune is implied by -march. And the |
| 26 |
docs say in the section about |
| 27 |
-mfpmath=sse: |
| 28 |
|
| 29 |
For the i386 compiler, you need to use `-march=CPU-TYPE', |
| 30 |
`-msse' or `-msse2' switches to enable SSE extensions and |
| 31 |
make this option effective. For the x86-64 compiler, these |
| 32 |
extensions are enabled by default. |
| 33 |
|
| 34 |
So it seems to me that -mfpmath does not need to be set. |
| 35 |
|
| 36 |
Well, I'd just use -march=native, unless when using distcc. If so, Daniel's |
| 37 |
tip about 'gcc -Q --help=target -march=native' will give you the exact |
| 38 |
options to use. |
| 39 |
|
| 40 |
Wonko |
Archives