[Piglit] [PATCH 1/8] arb_gpu_shader_fp64: Adds GS preprocessor define test
airlied at gmail.com
Thu Mar 17 20:50:34 UTC 2016
On 17 March 2016 at 22:20, Andres Gomez <agomez at igalia.com> wrote:
> On Thu, 2016-03-17 at 10:28 +1000, Dave Airlie wrote:
>> On 16 March 2016 at 01:50, Andres Gomez <agomez at igalia.com> wrote:
>> > Signed-off-by: Andres Gomez <agomez at igalia.com>
>> For completeness should we have tess and compute version of this test?
> In these series of additional tests we have focused in VS, GS and FS
> since most of the existing piglit tests are only considering those
That was because fp64 was written before tess/cs mostly, and I didn't have
access to any tess/cs to test on. Though I think we've done some tess
tests in the tess shader directory (not the fp64), again not sure where
things that are in two specs should land.
> In addition, we were creating the tests having in mind the status of
> development of the missing pieces of the i965 backend for this
> extension. Covering the VS, GS and FS stages suffices to test the two
> code paths used in the backend; scalar and vectorial. Hence,
> theoretically, and just for the case of i965, TESS and CS are also
> covered with the current tests.
> In any case, creating additional TESS and CS tests would be interesting,
> from an implementation agnostic point of view.
I'll try and look at the rest of the fp64 tests, they mostly pass on llvmpipe,
but I'd like to investigate the failures to see what fails.
More information about the Piglit