i965/fs: Improve accuracy of dFdy() to match dFdx().

Previously, we computed dFdy() using the following instruction:

  add(8) dst<1>F src<4,4,0)F -src.2<4,4,0>F { align1 1Q }

That had the disadvantage that it computed the same value for all 4
pixels of a 2x2 subspan, which meant that it was less accurate than
dFdx().  This patch changes it to the following instruction when
c->key.high_quality_derivatives is set:

  add(8) dst<1>F src<4,4,1>.xyxyF -src<4,4,1>.zwzwF { align16 1Q }

This gives it comparable accuracy to dFdx().

Unfortunately, align16 instructions can't be compressed, so in SIMD16
shaders, instead of emitting this instruction:

  add(16) dst<1>F src<4,4,1>.xyxyF -src<4,4,1>.zwzwF { align16 1H }

We need to unroll to two instructions:

  add(8) dst<1>F src<4,4,1>.xyxyF -src<4,4,1>.zwzwF { align16 1Q }
  add(8) (dst+1)<1>F (src+1)<4,4,1>.xyxyF -(src+1)<4,4,1>.zwzwF { align16 2Q }

Fixes piglit test spec/glsl-1.10/execution/fs-dfdy-accuracy.

Acked-by: Chris Forbes <[email protected]>
Reviewed-by: Eric Anholt <[email protected]>

0 files changed, 0 insertions, 0 deletions