fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-25 10:50:10 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

brw: Don't rely on SIMD splitting in opt_combine_convergent_txfs

Browse source 
The SIMD splitting pass does not handle wide force_writemask_all
instructions correctly at the moment.  For example, a SIMD32 TXF
on pre-Xe2 would get split to a pair of SIMD16.  But it will set
the groups to operate on channels 15:0 and 31:16.  That's not what
we want for a NoMask instruction - both should be 15:0, i.e.
bld.group(inst->exec_size, 0).

We could (and perhaps should) fix the SIMD splitting pass to handle
this, but the pass already has subtle complexity in which builders
are used.  Or we could alter fs_builder::group(), but that has broader
implications.  As a stop-gap, just make opt_combine_covergent_txfs stop
relying on SIMD splitting.  It's trivial to do and fixes the issue
without risking other breakage.

Fixes: 6341b3cd87 ("brw: Combine convergent texture buffer fetches into fewer loads")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32714>
This commit is contained in:
Kenneth Graunke 2024-12-18 22:41:52 -08:00 • committed by Marge Bot
parent 9b07cac02e
commit cb756ae8a2
1 changed files with 3 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

5
src/intel/compiler/brw_opt_txf_combiner.cpp
View file

@ -85,6 +85,7 @@ brw_opt_combine_convergent_txf(fs_visitor &s)
const def_analysis &defs = s.def_analysis.require();
const unsigned min_simd = 8 * reg_unit(s.devinfo);
const unsigned max_simd = 16 * reg_unit(s.devinfo);
const unsigned grf_size = REG_SIZE * reg_unit(s.devinfo);
bool progress = false;
@ -148,8 +149,8 @@ brw_opt_combine_convergent_txf(fs_visitor &s)
continue;
/* Emit divergent TXFs and replace the original ones with MOVs */
for (unsigned curr = 0; curr < count; curr += 32) {
const unsigned lanes = CLAMP(count - curr, min_simd, 32);
for (unsigned curr = 0; curr < count; curr += max_simd) {
const unsigned lanes = CLAMP(count - curr, min_simd, max_simd);
const unsigned width = util_next_power_of_two(lanes);
const fs_builder ubld =
fs_builder(&s).at(block, txfs[curr]).exec_all().group(width, 0);