[Piglit] [PATCH] spec/arb_blend_func_extended: Add test for SIMD16 dual source blending

Matt Turner mattst88 at gmail.com
Thu Mar 5 10:51:20 PST 2015

On Thu, Mar 5, 2015 at 10:50 AM, Matt Turner <mattst88 at gmail.com> wrote:
> From: Iago Toral Quiroga <itoral at igalia.com>
> On Intel hardware at least, SIMD16 dual source rendering requires handling
> pixel data in two sets of 8 pixels each. Incorrect implementations may fail
> to map correct colors for each pixel group (for example by using the color
> for the first group as the color for the second group or viceversa). However,
> tests that render using solid colors across the entire polygon won't catch
> these cases (since in that case the color is the same for boths groups of
> pixels).
> This test blends using a checker board pattern where each cell is
> 10px wide and 10px tall. This makes it so that the two sets of 8 pixels
> issued during SIMD16 operation pack different color data for the pixels
> involved, enabling testing of correct behavior in that case.
> ---
> This is the same patch Iago sent, just rebased on today's all.py changes.
> Iago, the test doesn't seem to open a window. I expected it would render
> to an offscreen FBO but then blit to a window to display if -auto wasn't
> passed.

Ugh, I apparently used the message-id of the corresponding Mesa patch
in reply-to instead of the piglit test.

More information about the Piglit mailing list