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: fix opquantize2f16 on GFX6-7

Browse source 
Make sure to preserve signed zeroes.

Fixes dEQP-VK.spirv_assembly.instruction.compute.opquantize.flush_to_zero
on GFX6 (Pitcairn). Untested on GFX7.

Fixes: 54a09545ec ("aco: optimize a*0.0")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10319>
(cherry picked from commit 9434675d60)
This commit is contained in:
Samuel Pitoiset 2021-04-19 13:47:14 +02:00 • committed by Eric Engestrom
parent 9d518272a6
commit af0926bac0
2 changed files with 6 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.pick_status.json
View file

@ -841,7 +841,7 @@
"description": "aco: fix opquantize2f16 on GFX6-7",
"nominated": true,
"nomination_type": 1,
"resolution": 0,
"resolution": 1,
"master_sha": null,
"because_sha": "54a09545ec639e9204e6ee823e52872897e793df"
},

9
src/amd/compiler/aco_instruction_selection.cpp
View file

@ -3056,12 +3056,13 @@ void visit_alu_instr(isel_context *ctx, nir_alu_instr *instr)
*/
f32 = bld.vop1(aco_opcode::v_cvt_f32_f16, bld.def(v1), f16);
Temp smallest = bld.copy(bld.def(s1), Operand(0x38800000u));
Instruction* vop3 = bld.vopc_e64(aco_opcode::v_cmp_nlt_f32, bld.hint_vcc(bld.def(bld.lm)), f32, smallest);
vop3->vop3().abs[0] = true;
cmp_res = vop3->definitions[0].getTemp();
Instruction* tmp0 = bld.vopc_e64(aco_opcode::v_cmp_lt_f32, bld.def(bld.lm), f32, smallest);
tmp0->vop3().abs[0] = true;
Temp tmp1 = bld.vopc(aco_opcode::v_cmp_lg_f32, bld.hint_vcc(bld.def(bld.lm)), Operand(0u), f32);
cmp_res = bld.sop2(aco_opcode::s_nand_b64, bld.def(s2), bld.def(s1, scc), tmp0->definitions[0].getTemp(), tmp1);
}
if (ctx->block->fp_mode.preserve_signed_zero_inf_nan32 || ctx->program->chip_class < GFX8) {
if (ctx->block->fp_mode.preserve_signed_zero_inf_nan32) {
Temp copysign_0 = bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), Operand(0u), as_vgpr(ctx, src));
bld.vop2(aco_opcode::v_cndmask_b32, Definition(dst), copysign_0, f32, cmp_res);
} else {