| 1 |
On Thu, Feb 12, 2009 at 12:58 AM, Neil Bothwick <neil@××××××××××.uk> wrote: |
| 2 |
> On Wed, 11 Feb 2009 20:22:32 -0800, Mark Knecht wrote: |
| 3 |
> |
| 4 |
>> > CFLAGS? The machine that fails: |
| 5 |
>> > |
| 6 |
>> > CFLAGS="-O3 -march=athlon-xp -funroll-loops -fprefetch-loop-arrays -pipe" |
| 7 |
> |
| 8 |
>> CFLAGS="-O2 -march=i686 -fomit-frame-pointer -pipe" |
| 9 |
> |
| 10 |
> You have more aggressive flags on the failing machine, have you tried |
| 11 |
> using the more conservative (less ricey) flags? |
| 12 |
> |
| 13 |
> |
| 14 |
> -- |
| 15 |
> Neil Bothwick |
| 16 |
|
| 17 |
I have not and intend to do that today. The flags have been set like |
| 18 |
this since 2003 so I'm surprised that it might be something like this |
| 19 |
but it's clearly possible. |
| 20 |
|
| 21 |
I'm unqualified to say what part of the differences is making this set |
| 22 |
more aggressive. Line by line what's your opinion? |
| 23 |
|
| 24 |
-O3 vs -O2 ? |
| 25 |
-march=athlon-xp vs -march=i686 |
| 26 |
-funroll-loops |
| 27 |
-fprefetch-loop-arrays |
| 28 |
not using -fomit-frame-pointer |
| 29 |
|
| 30 |
I would think that I should be able to also make the flags more |
| 31 |
aggressive on a passing machine and see the same failure, assuming |
| 32 |
this is the root cause. |
| 33 |
|
| 34 |
Strange taht after 6-years this would come up but stranger thing happen. |
| 35 |
|
| 36 |
Thanks, |
| 37 |
Mark |
Archives