fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-24 17:30:12 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

radv/nir/lower_cmat: clean up gfx12 transpose

Browse source 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37213>
This commit is contained in:
Georg Lehmann 2025-09-06 17:03:48 +02:00 • committed by Marge Bot
parent 2da7b4bd0a
commit 5c0ebcdaef
1 changed files with 6 additions and 10 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
src/amd/vulkan/nir/radv_nir_lower_cooperative_matrix.c
View file

@ -607,21 +607,17 @@ convert_use(nir_builder *b, nir_def *src, enum glsl_cmat_use src_use, enum glsl_
assert(num_comps == 16 || params->gfx_level >= GFX12);
if (params->gfx_level >= GFX12) {
if (params->wave_size == 64) {
nir_def *cond = nir_inverse_ballot_imm(b, 0xf0f0f0f00f0f0f0f, params->wave_size);
/* One component contains 2/4 rows in wave32/64, so we must transpose inside it. */
for (int cross32 = params->wave_size == 64; cross32 >= 0; cross32--) {
uint64_t even = cross32 ? 0xf0f0f0f00f0f0f0f : 0xff0000ffff0000ff;
nir_def *cond = nir_inverse_ballot_imm(b, even, params->wave_size);
unsigned x_mask = cross32 ? 0x24 : 0x18;
for (unsigned i = 0; i < num_comps; i++) {
nir_def *comp = components[i];
nir_def *compx = shuffle_xor_imm(b, comp, 0x24);
nir_def *compx = shuffle_xor_imm(b, comp, x_mask);
components[i] = nir_bcsel(b, cond, comp, compx);
}
}
nir_def *cond = nir_inverse_ballot_imm(b, 0xff0000ffff0000ff, params->wave_size);
for (unsigned i = 0; i < num_comps; i++) {
nir_def *comp = components[i];
nir_def *compx = shuffle_xor_imm(b, comp, 0x18);
components[i] = nir_bcsel(b, cond, comp, compx);
}
}
}