fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-03-18 21:00:33 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

aco/ra: create v_dot2c_f32_f16

Browse source 
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40003>
This commit is contained in:
Georg Lehmann 2026-01-26 19:24:48 +01:00 • committed by Marge Bot
parent 237b8ca205
commit 3238e64d3c
2 changed files with 7 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

3
src/amd/compiler/aco_ir.cpp
View file

@ -1492,7 +1492,8 @@ get_tied_defs(Instruction* instr)
instr->opcode == aco_opcode::v_fmac_legacy_f32 ||
instr->opcode == aco_opcode::v_pk_fmac_f16 || instr->opcode == aco_opcode::v_writelane_b32 ||
instr->opcode == aco_opcode::v_writelane_b32_e64 ||
instr->opcode == aco_opcode::v_dot4c_i32_i8 || instr->opcode == aco_opcode::s_fmac_f32 ||
instr->opcode == aco_opcode::v_dot4c_i32_i8 ||
instr->opcode == aco_opcode::v_dot2c_f32_f16 || instr->opcode == aco_opcode::s_fmac_f32 ||
instr->opcode == aco_opcode::s_fmac_f16) {
ops.push_back(2);
} else if (instr->opcode == aco_opcode::s_addk_i32 || instr->opcode == aco_opcode::s_mulk_i32 ||

5
src/amd/compiler/aco_register_allocation.cpp
View file

@ -3127,6 +3127,10 @@ vop3_can_use_vop2acc(ra_ctx& ctx, Instruction* instr)
if (!ctx.program->dev.has_fmac_legacy32)
return false;
break;
case aco_opcode::v_dot2_f32_f16:
if (ctx.program->gfx_level < GFX10 || ctx.program->gfx_level >= GFX12)
return false;
break;
default: return false;
}
@ -3537,6 +3541,7 @@ optimize_encoding_vop2(ra_ctx& ctx, RegisterFile& register_file, aco_ptr<Instruc
case aco_opcode::v_dot4_i32_i8: instr->opcode = aco_opcode::v_dot4c_i32_i8; break;
case aco_opcode::v_mad_legacy_f32: instr->opcode = aco_opcode::v_mac_legacy_f32; break;
case aco_opcode::v_fma_legacy_f32: instr->opcode = aco_opcode::v_fmac_legacy_f32; break;
case aco_opcode::v_dot2_f32_f16: instr->opcode = aco_opcode::v_dot2c_f32_f16; break;
default: break;
}
}