This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] Fix PR37686: Correct how lambda gets the loop index for the matrix representation of array accesses
On Wed, Oct 8, 2008 at 9:09 PM, Sebastian Pop <sebpop@gmail.com> wrote:
> On Fri, Oct 3, 2008 at 6:49 PM, Andrew Pinski <pinskia@gmail.com> wrote:
>>>> +/* { dg-options "-mcpu=power4 -m64 -mtune=cell -O3 -ftree-loop-linear --param ggc-min-expand=0 --param ggc-min-heapsize=0 -funroll-loops -fpreprocessed" } */
>>>
>>> Either the testcase is ppc specific and then it should go under
>>> gcc.target/powerpc/, or it is not and then you just should compile
>>> it everywhere, perhaps with special flags on ppc and just use
>>> -O3 -ftree-loop-linear -funroll-loops on all the other arches.
>>> My preference would be the latter in this case.
>>
>> In fact my options were only needed to get the crash happening as it
>> was very memory layout dependent; I don't think we need all of these
>> special options really.
>>
>
> What about using these flags:
>
> /* { dg-options "-O3 -ftree-loop-linear" } */
>
> Okay for trunk with this change? The patch passed regstrap on
> amd64-linux.
Ok.
Thanks,
Richard.