AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

aco/optimizer: use new helpers for add16 opts

Browse Source 
Foz-DB Navi48:
Totals from 164 (0.20% of 82419) affected shaders:
Instrs: 145304 -> 145335 (+0.02%); split: -0.00%, +0.02%
CodeSize: 794156 -> 794280 (+0.02%); split: -0.00%, +0.02%
Latency: 1884349 -> 1884227 (-0.01%); split: -0.01%, +0.00%
InvThroughput: 350403 -> 350393 (-0.00%)

Foz-DB Navi21:
Totals from 164 (0.20% of 82387) affected shaders:
Instrs: 117416 -> 117414 (-0.00%)
CodeSize: 673328 -> 673312 (-0.00%)
Latency: 1896952 -> 1897094 (+0.01%); split: -0.00%, +0.01%
InvThroughput: 638536 -> 638556 (+0.00%); split: -0.01%, +0.01%
Copies: 14579 -> 14577 (-0.01%)
VALU: 65895 -> 65893 (-0.00%)

Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38150>
This commit is contained in:
Georg Lehmann
2024-12-14 18:10:39 +01:00
committed by Marge Bot
parent e8f5b9374b
commit fec10ea3ea
1 changed files with 7 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/amd/compiler/aco_optimizer.cpp
View File

@@ -4788,14 +4788,6 @@ combine_instruction(opt_ctx& ctx, aco_ptr<Instruction>& instr)
}
} else if (instr->opcode == aco_opcode::v_not_b32 && ctx.program->gfx_level >= GFX10) {
combine_not_xor(ctx, instr);
} else if (instr->opcode == aco_opcode::v_add_u16 && !instr->valu().clamp) {
combine_three_valu_op(
ctx, instr, aco_opcode::v_mul_lo_u16,
ctx.program->gfx_level == GFX8 ? aco_opcode::v_mad_legacy_u16 : aco_opcode::v_mad_u16,
"120", 1 | 2);
} else if (instr->opcode == aco_opcode::v_add_u16_e64 && !instr->valu().clamp) {
combine_three_valu_op(ctx, instr, aco_opcode::v_mul_lo_u16_e64, aco_opcode::v_mad_u16, "120",
1 | 2);
} else if (instr->opcode == aco_opcode::v_add_u32 && !instr->usesModifiers()) {
if (combine_add_sub_b2i(ctx, instr, aco_opcode::v_addc_co_u32, 1 | 2)) {
} else if (combine_add_bcnt(ctx, instr)) {
@@ -5008,6 +5000,13 @@ combine_instruction(opt_ctx& ctx, aco_ptr<Instruction>& instr)
/* v_mul_f32(a, v_cndmask_b32(0, 1.0, cond)) -> v_cndmask_b32(0, a, cond) */
add_opt(v_cndmask_b32, v_cndmask_b32, 0x3, "1032",
and_cb<check_const_cb<0, 0>, remove_const_cb<0x3f800000>>, true);
} else if (info.opcode == aco_opcode::v_add_u16 && !info.clamp) {
if (ctx.program->gfx_level < GFX9)
add_opt(v_mul_lo_u16, v_mad_legacy_u16, 0x3, "120");
else
add_opt(v_mul_lo_u16, v_mad_u16, 0x3, "120");
} else if (info.opcode == aco_opcode::v_add_u16_e64 && !info.clamp) {
add_opt(v_mul_lo_u16_e64, v_mad_u16, 0x3, "120");
}
if (match_and_apply_patterns(ctx, info, patterns)) {