mirror of
https://gitlab.freedesktop.org/mesa/mesa.git
synced 2026-05-04 20:38:06 +02:00
i965/fs: No need to unzip SIMD-periodic sources during SIMD lowering.
If the source value is going to the same for all SIMD-lowered chunks of the instruction there should be no need to unzip the value into multiple temporary registers one for each lowered chunk. As a side effect this fixes SIMD lowering of instructions with a vector immediate source. In the long term it *might* still be worth fixing offset() to handle vector immediates correctly though, this should be good enough for the moment. Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
This commit is contained in:
parent
168163f5f0
commit
1760c24b4b
1 changed files with 1 additions and 1 deletions
|
|
@ -4767,7 +4767,7 @@ fs_visitor::lower_simd_width()
|
|||
|
||||
for (unsigned j = 0; j < inst->sources; j++) {
|
||||
if (inst->src[j].file != BAD_FILE &&
|
||||
!is_uniform(inst->src[j])) {
|
||||
!is_periodic(inst->src[j], lower_width)) {
|
||||
/* Get the i-th copy_width-wide chunk of the source. */
|
||||
const fs_builder cbld = lbld.group(copy_width, 0);
|
||||
const fs_reg src = offset(inst->src[j], cbld, i);
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue