[PATCH 1/9] [nvptx] Enable large vectors

[Thomas Schwinge]

Hi!

On 2019-01-12T23:21:23+0100, Tom de Vries <tdevries@suse.de> wrote:
> Allow vector_length clauses to accept values larger than warp size.

>       * testsuite/libgomp.oacc-c-c++-common/parallel-dims.c: Expect vector
>       length 2097152 to be reduced to 1024 instead of 32.

> --- a/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
> +++ b/libgomp/testsuite/libgomp.oacc-c-c++-common/parallel-dims.c
> @@ -350,7 +350,7 @@ int main ()
>      int gangs_min, gangs_max, workers_min, workers_max, vectors_min, vectors_max;
>      gangs_min = workers_min = vectors_min = INT_MAX;
>      gangs_max = workers_max = vectors_max = INT_MIN;
> -#pragma acc parallel copy (vectors_actual) /* { dg-warning "using vector_length \\(32\\), ignoring 2097152" "" { target openacc_nvidia_accel_configured } } */ \
> +#pragma acc parallel copy (vectors_actual) /* { dg-warning "using vector_length \\(1024\\), ignoring 2097152" "" { target openacc_nvidia_accel_configured } } */ \
>    vector_length (VECTORS)
>      {
>        if (acc_on_device (acc_device_host))
> @@ -361,7 +361,7 @@ int main ()
>        else if (acc_on_device (acc_device_nvidia))
>       {
>         /* The GCC nvptx back end enforces vector_length (32).  */
> -       vectors_actual = 32;
> +       vectors_actual = 1024;
>       }
>        else
>       __builtin_abort ();

As obvious, pushed "[nvptx] Update comment in
'libgomp.oacc-c-c++-common/parallel-dims.c'" to master branch in commit
e64d62c7008e6a4b0227fd25e071db8f0b3f1820, see attached.


Grüße
 Thomas


-----------------
Mentor Graphics (Deutschland) GmbH, Arnulfstrasse 201, 80634 München Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, Frank Thürauf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-nvptx-Update-comment-in-libgomp.oacc-c-c-common-para.patch
Type: text/x-diff
Size: 1325 bytes
Desc: not available
URL: <https://gcc.gnu.org/pipermail/gcc-patches/attachments/20210608/a384d0b3/attachment.bin>