GitLab

I have observed some shaders in shader-db that have code like:

void main ()
{
  movieTaps[0] = vec3(0.2165, 0.125, 1.0);
  // repeat with different constant vec3 for [1..58].
  movieTaps[59] = vec3(0.8668, 0.2513, 0.907);

And NIR merrily generates the obvious, horrifying code. The worst part is that since these are vec3, each component is written individually.

        vec1 32 ssa_0 = load_const (0x00000000 = 0.000000)
        vec1 32 ssa_1 = load_const (0x3f800000 = 1.000000)
        ...
        vec1 32 ssa_10 = load_const (0x3e5db22d = 0.216500)
        intrinsic store_scratch (ssa_10, ssa_0) (align_mul=256, align_offset=0, wrmask=x /*1*/)
        vec1 32 ssa_11 = load_const (0x3e000000 = 0.125000)
        vec1 32 ssa_12 = load_const (0x00000004 = 0.000000)
        intrinsic store_scratch (ssa_11, ssa_12) (align_mul=256, align_offset=0, wrmask=x /*1*/)
        vec1 32 ssa_13 = load_const (0x00000008 = 0.000000)
        intrinsic store_scratch (ssa_1, ssa_13) (align_mul=256, align_offset=0, wrmask=x /*1*/)

The Intel compiler generates the obvious, horrifying code. As a side note, I noticed this because those writes get misidentified as 180 spills.

Since we have code that converts constant arrays into uniforms, it should be possible to detect this case and convert it to a constant array. I think such a pass would progress like:

1. Find all writes to the array that dominate all reads. Convert those writes to be part of the initializer for the array. Care must be taken so that multiple writes to the same location would be processed in the correct order.

2. If a non-const array has only reads, make it const.

It's possible that the app with this weird array initialization may have changed since it was captured to shader-db, but the issue of unmerged scratch writes is more likely to affect other shaders. I'll open a separate issue for that.