fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-09 02:28:10 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

aco/optimizer: optimize packed fneg with negative constant

Browse source 
We shouldn't rely on constant propagation always creating
positive 1.0 constants.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35020>
This commit is contained in:
Georg Lehmann 2024-10-21 18:59:05 +02:00 • committed by Marge Bot
parent fa3f207035
commit 7d59bd399a
1 changed files with 9 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
src/amd/compiler/aco_optimizer.cpp
View file

@ -3323,11 +3323,14 @@ combine_vop3p(opt_ctx& ctx, aco_ptr<Instruction>& instr)
ssa_info& info = ctx.info[op.tempId()];
if (info.parent_instr->opcode == aco_opcode::v_pk_mul_f16 &&
(info.parent_instr->operands[0].constantEquals(0x3C00) ||
info.parent_instr->operands[1].constantEquals(0x3C00))) {
info.parent_instr->operands[1].constantEquals(0x3C00) ||
info.parent_instr->operands[0].constantEquals(0xBC00) ||
info.parent_instr->operands[1].constantEquals(0xBC00))) {
VALU_instruction* fneg = &info.parent_instr->valu();
unsigned fneg_src = fneg->operands[0].constantEquals(0x3C00);
unsigned fneg_src =
fneg->operands[0].constantEquals(0x3C00) || fneg->operands[0].constantEquals(0xBC00);
if (fneg->opsel_lo[1 - fneg_src] || fneg->opsel_hi[1 - fneg_src])
continue;
@ -3351,6 +3354,10 @@ combine_vop3p(opt_ctx& ctx, aco_ptr<Instruction>& instr)
bool opsel_hi = vop3p->opsel_hi[i];
bool neg_lo = fneg->neg_lo[0] ^ fneg->neg_lo[1];
bool neg_hi = fneg->neg_hi[0] ^ fneg->neg_hi[1];
bool neg_const = fneg->operands[1 - fneg_src].constantEquals(0xBC00);
/* Avoid ternary xor as it causes CI fails that can't be reproduced on other systems. */
neg_lo ^= neg_const;
neg_hi ^= neg_const;
vop3p->neg_lo[i] ^= opsel_lo ? neg_hi : neg_lo;
vop3p->neg_hi[i] ^= opsel_hi ? neg_hi : neg_lo;
vop3p->opsel_lo[i] ^= opsel_lo ? !fneg->opsel_hi[fneg_src] : fneg->opsel_lo[fneg_src];