| 1 |
Am 22.07.2011 00:35, schrieb walt: |
| 2 |
> On 07/20/2011 10:54 PM, Bill Kenworthy wrote: |
| 3 |
> |
| 4 |
>> So not sure about march=native now as it is only what was built with |
| 5 |
>> native thats been problematic. |
| 6 |
> |
| 7 |
> Makes me wonder if gcc and glibc need to be recompiled with arch=native |
| 8 |
> before rebuilding the rest of the system? |
| 9 |
> |
| 10 |
> |
| 11 |
> |
| 12 |
|
| 13 |
No. See my posts in the other part of this thread where threading was |
| 14 |
once broken by a stupid mobile phone MUA. |
| 15 |
|
| 16 |
Every "native" setting is just resolved by gcc at compile time to some |
| 17 |
concrete setting like "core2". There is nothing magical about it and you |
| 18 |
also don't need to recompile glibc or gcc any more than you need after |
| 19 |
switching your march settings between two other settings. |
| 20 |
|
| 21 |
No one had any breakage during the time when -march=core2 was introduced |
| 22 |
and all Core2 owners switched from -march=nocona to -march=core2. There |
| 23 |
should be no more breakage when switching to "native". |
| 24 |
|
| 25 |
Correlation does not imply causation. |
| 26 |
|
| 27 |
Regards, |
| 28 |
Florian Philipp |
Archives