[Piglit] gl_PointSize, vs-point_size-zero test

Tapani Pälli tapani.palli at intel.com
Tue Sep 22 02:55:13 PDT 2015

On 09/22/2015 12:46 PM, Marek Olšák wrote:
> The point size clamping should take place after transform feedback,
> i.e. not in the shader.

OK, so I'm doing it too early. Seems strange to allow transferring 
values from stage to another that will eventually get clamped but I 
guess that's how it is. Will try to clamp later.

> Marek
> On Tue, Sep 22, 2015 at 5:20 AM, Tapani Pälli <tapani.palli at intel.com> wrote:
>> Hi;
>> I noticed that i965 driver currently fails vs-point-size-zero test.
>> I have a fix for this where I clamp the value in driver for min and max, as
>> specified by the OpenGL spec (using values in GL_POINT_SIZE_RANGE) which
>> makes this test pass.
>> However with the fix I noticed 3 regressing tests:
>> ext_transform_feedback-builtin-varyings gl_PointSize -auto
>> glsl-1.50/execution/gs-redeclares-pervertex-out-only.shader_test
>> glsl-1.50/execution/gs-redeclares-both-pervertex-blocks.shader_test
>> It looks like all of these assume that gl_PointSize is not clamped but will
>> preserve values less than zero in the shader.
>> In your opinion, are these tests OK or should test be fixed to not assume
>> having values less than 1.0 for gl_Pointsize?
>> Thanks;
>> // Tapani

More information about the Piglit mailing list