fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-07 09:18:04 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

aco/ra: refactor get_reg_simple() with increased stride.

Browse source 
This should avoid some redundant calls.

Totals from 153 (0.19% of 79395) affected shaders: (GFX11)

Instrs: 301717 -> 301687 (-0.01%); split: -0.06%, +0.05%
CodeSize: 1583080 -> 1582988 (-0.01%); split: -0.06%, +0.05%
VGPRs: 10068 -> 10348 (+2.78%)
Latency: 6685446 -> 6685475 (+0.00%); split: -0.11%, +0.11%
InvThroughput: 999241 -> 999316 (+0.01%); split: -0.01%, +0.02%
VClause: 3868 -> 3870 (+0.05%)
Copies: 23752 -> 23769 (+0.07%); split: -0.27%, +0.34%
Branches: 6479 -> 6480 (+0.02%)
VALU: 179290 -> 179307 (+0.01%); split: -0.04%, +0.04%
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29235>
This commit is contained in:
Daniel Schürmann 2024-05-14 16:20:36 +02:00 • committed by Marge Bot
parent 1b0edf3f33
commit 8e817cf52b
1 changed files with 8 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
src/amd/compiler/aco_register_allocation.cpp
View file

@ -900,15 +900,14 @@ get_reg_simple(ra_ctx& ctx, const RegisterFile& reg_file, DefInfo info)
uint32_t stride = info.rc.is_subdword() ? DIV_ROUND_UP(info.stride, 4) : info.stride;
RegClass rc = info.rc;
DefInfo new_info = info;
new_info.rc = RegClass(rc.type(), size);
for (unsigned new_stride = 16; new_stride > stride; new_stride /= 2) {
if (size % new_stride)
continue;
new_info.stride = new_stride;
std::optional<PhysReg> res = get_reg_simple(ctx, reg_file, new_info);
if (res)
return res;
if (stride < size && !rc.is_subdword()) {
DefInfo new_info = info;
new_info.stride = stride * 2;
if (size % new_info.stride == 0) {
std::optional<PhysReg> res = get_reg_simple(ctx, reg_file, new_info);
if (res)
return res;
}
}
auto is_free = [&](PhysReg reg_index)