fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-01-03 09:20:13 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

aco: Combine s_abs and s_sub/s_add to s_absdiff.

Browse source 
Totals from 2 (0.00% of 134913) affected shaders:
CodeSize: 1344 -> 1336 (-0.60%)
Instrs: 277 -> 275 (-0.72%)

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18870>
This commit is contained in:
Georg Lehmann 2022-09-23 19:00:36 +02:00 • committed by Marge Bot
parent 7e1d77fd90
commit 853d2cb6f1
1 changed files with 40 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

40
src/amd/compiler/aco_optimizer.cpp
View file

@ -2772,6 +2772,44 @@ combine_salu_lshl_add(opt_ctx& ctx, aco_ptr<Instruction>& instr)
return false;
}
/* s_abs_i32(s_sub_[iu]32(a, b)) -> s_absdiff_i32(a, b)
* s_abs_i32(s_add_[iu]32(a, #b)) -> s_absdiff_i32(a, -b)
*/
bool
combine_sabsdiff(opt_ctx& ctx, aco_ptr<Instruction>& instr)
{
if (!instr->operands[0].isTemp() || !ctx.info[instr->operands[0].tempId()].is_add_sub())
return false;
Instruction* op_instr = follow_operand(ctx, instr->operands[0], false);
if (!op_instr)
return false;
if (op_instr->opcode == aco_opcode::s_add_i32 || op_instr->opcode == aco_opcode::s_add_u32) {
for (unsigned i = 0; i < 2; i++) {
uint64_t constant;
if (op_instr->operands[!i].isLiteral() ||
!is_operand_constant(ctx, op_instr->operands[i], 32, &constant))
continue;
if (op_instr->operands[i].isTemp())
ctx.uses[op_instr->operands[i].tempId()]--;
op_instr->operands[0] = op_instr->operands[!i];
op_instr->operands[1] = Operand::c32(-int32_t(constant));
goto use_absdiff;
}
return false;
}
use_absdiff:
op_instr->opcode = aco_opcode::s_absdiff_i32;
std::swap(instr->definitions[0], op_instr->definitions[0]);
std::swap(instr->definitions[1], op_instr->definitions[1]);
ctx.uses[instr->operands[0].tempId()]--;
return true;
}
bool
combine_add_sub_b2i(opt_ctx& ctx, aco_ptr<Instruction>& instr, aco_opcode new_op, uint8_t ops)
{
@ -4186,6 +4224,8 @@ combine_instruction(opt_ctx& ctx, aco_ptr<Instruction>& instr)
} else if (combine_constant_comparison_ordering(ctx, instr)) {
} else if (combine_salu_n2(ctx, instr)) {
}
} else if (instr->opcode == aco_opcode::s_abs_i32) {
combine_sabsdiff(ctx, instr);
} else if (instr->opcode == aco_opcode::v_and_b32) {
combine_and_subbrev(ctx, instr);
} else if (instr->opcode == aco_opcode::v_fma_f32 || instr->opcode == aco_opcode::v_fma_f16) {