OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

intel/fs_reg_allocate: Improve compressed instruction self-interference

Browse Source 
The old version worked fine for SIMD16 instructions but SIMD8
instructions where the destination spans two registers have the same
problem.

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17908>
This commit is contained in:
Jason Ekstrand
2020-09-07 01:21:23 -05:00
committed by Marge Bot
parent 1ce5be916f
commit caed8df146
1 changed files with 11 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
src/intel/compiler/brw_fs_reg_allocate.cpp
View File

@@ -531,18 +531,18 @@ fs_reg_alloc::setup_inst_interference(const fs_inst *inst)
} }
} }
/* In 16-wide instructions we have an issue where a compressed /* A compressed instruction is actually two instructions executed
* instruction is actually two instructions executed simultaneously. * simultaneously. On most platforms, it ok to have the source and
* It's actually ok to have the source and destination registers be * destination registers be the same. In this case, each instruction
* the same. In this case, each instruction over-writes its own * over-writes its own source and there's no problem. The real problem
* source and there's no problem. The real problem here is if the * here is if the source and destination registers are off by one. Then
* source and destination registers are off by one. Then you can end * you can end up in a scenario where the first instruction over-writes the
* up in a scenario where the first instruction over-writes the * source of the second instruction. Since the compiler doesn't know about
* source of the second instruction. Since the compiler doesn't know * this level of granularity, we simply make the source and destination
* about this level of granularity, we simply make the source and * interfere.
* destination interfere.
*/ */
if (inst->exec_size >= 16 && inst->dst.file == VGRF) { if (inst->dst.component_size(inst->exec_size) > REG_SIZE &&
inst->dst.file == VGRF) {
for (int i = 0; i < inst->sources; ++i) { for (int i = 0; i < inst->sources; ++i) {
if (inst->src[i].file == VGRF) { if (inst->src[i].file == VGRF) {
ra_add_node_interference(g, first_vgrf_node + inst->dst.nr, ra_add_node_interference(g, first_vgrf_node + inst->dst.nr,