/third_party/mesa3d/src/amd/compiler/tests/ |
D | test_optimizer.cpp | 37 writeout(0, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], neg_b)); 44 writeout(1, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), Operand::c32(0x123456u), neg_a)); 49 writeout(2, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), neg_neg_a, inputs[1])); 54 writeout(3, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), abs_neg_a, inputs[1])); 60 writeout(4, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), neg_abs_a, inputs[1])); 68 writeout(6, bld.vop2(aco_opcode::v_add_f32, bld.def(v1), neg_a, inputs[1])); 72 writeout(7, bld.vop2(aco_opcode::v_add_f32, bld.def(v1), inputs[1], neg_a)); 77 writeout(8, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], neg_c)); 82 writeout(9, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), abs_neg_abs_a, inputs[1])); 99 Temp tmp = bld.vop2(aco_opcode::v_add_f32, bld.def(v1), inputs[0], inputs[1]); [all …]
|
D | test_sdwa.cpp | 173 writeout(0, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfe_byte0_b)); 178 writeout(1, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfe_byte1_b)); 183 writeout(2, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfe_byte2_b)); 188 writeout(3, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfe_byte3_b)); 193 writeout(4, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfe_word0_b)); 198 writeout(5, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfe_word1_b)); 202 writeout(6, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfi_byte0_b)); 207 writeout(7, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfi_word0_b)); 219 writeout(8, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], bfi_byte1_b)); 290 writeout(0, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), inputs[0], neg_byte0)); [all …]
|
D | test_optimizer_postRA.cpp | 301 Temp res0 = bld.vop2(aco_opcode::v_add_f32, bld.def(v1, reg_v2), Operand(tmp0, reg_v2), b); 308 Temp res1 = bld.vop2(aco_opcode::v_sub_f32, bld.def(v1, reg_v2), b, Operand(tmp1, reg_v2)); 323 Temp res3 = bld.vop2(aco_opcode::v_add_f32, bld.def(v1, reg_v2), Operand(tmp3, reg_v2), b); 360 … Temp res8 = bld.vop2(aco_opcode::v_cndmask_b32, bld.def(v1, reg_v2), Operand(tmp8, reg_v2), b, c); 367 … Temp res9 = bld.vop2(aco_opcode::v_cndmask_b32, bld.def(v1, reg_v2), Operand(tmp9, reg_v2), b, d); 383 Temp res10 = bld.vop2(aco_opcode::v_add_f32, bld.def(v1, reg_v2), Operand(tmp10, reg_v2), b); 393 Temp res11 = bld.vop2(aco_opcode::v_add_f32, bld.def(v1, reg_v2), Operand(tmp11_1, reg_v2), b);
|
D | helpers.cpp | 274 return b.vop2(aco_opcode::v_mul_f16, b.def(v2b), Operand::c16(0xbc00u), src); in fneg() 276 return b.vop2(aco_opcode::v_mul_f32, b.def(v1), Operand::c32(0xbf800000u), src); in fneg() 310 return b.vop2(aco_opcode::v_add_f16, b.def(v2b), src0, src1); in fadd() 312 return b.vop2(aco_opcode::v_add_f32, b.def(v1), src0, src1); in fadd() 318 return b.vop2(aco_opcode::v_mul_f16, b.def(v2b), src0, src1); in fmul() 320 return b.vop2(aco_opcode::v_mul_f32, b.def(v1), src0, src1); in fmul()
|
/third_party/mesa3d/src/amd/compiler/ |
D | aco_lower_to_hw_instr.cpp | 259 bld.vop2(aco_opcode::v_cndmask_b32, dst[0], vtmp_op[0], src1[0], Operand(vcc, bld.lm)); in emit_int64_dpp_op() 260 bld.vop2(aco_opcode::v_cndmask_b32, dst[1], vtmp_op[1], src1[1], Operand(vcc, bld.lm)); in emit_int64_dpp_op() 329 bld.vop2(aco_opcode::v_add_co_u32, dst[0], bld.def(bld.lm, vcc), src0[0], src1[0]); in emit_int64_op() 331 bld.vop2(aco_opcode::v_addc_co_u32, dst[1], bld.def(bld.lm, vcc), src0[1], src1[1], in emit_int64_op() 334 bld.vop2(aco_opcode::v_and_b32, dst[0], src0[0], src1[0]); in emit_int64_op() 335 bld.vop2(aco_opcode::v_and_b32, dst[1], src0[1], src1[1]); in emit_int64_op() 337 bld.vop2(aco_opcode::v_or_b32, dst[0], src0[0], src1[0]); in emit_int64_op() 338 bld.vop2(aco_opcode::v_or_b32, dst[1], src0[1], src1[1]); in emit_int64_op() 340 bld.vop2(aco_opcode::v_xor_b32, dst[0], src0[0], src1[0]); in emit_int64_op() 341 bld.vop2(aco_opcode::v_xor_b32, dst[1], src0[1], src1[1]); in emit_int64_op() [all …]
|
D | aco_instruction_selection.cpp | 164 return bld.vop2(aco_opcode::v_mbcnt_hi_u32_b32, Definition(dst), mask_hi, mbcnt_lo); in emit_mbcnt() 213 Operand index_x4 = bld.vop2(aco_opcode::v_lshlrev_b32, bld.def(v1), Operand::c32(2u), index); in emit_bpermute() 228 Temp index_x4 = bld.vop2(aco_opcode::v_lshlrev_b32, bld.def(v1), Operand::c32(2u), index); in emit_bpermute() 295 bld.vop2(aco_opcode::v_lshrrev_b32, Definition(dst), Operand::c32(util_logbase2(b)), a); in emit_v_div_u32() 316 bld.vop2(aco_opcode::v_lshrrev_b32, Definition(pre_shift_dst), Operand::c32(info.pre_shift), in emit_v_div_u32() 334 bld.vop2(aco_opcode::v_lshrrev_b32, Definition(dst), Operand::c32(info.post_shift), in emit_v_div_u32() 687 Temp high = bld.vop2(aco_opcode::v_ashrrev_i32, bld.def(v1), Operand::c32(31u), tmp); in convert_int() 901 Temp tmp = bld.vop2(opc, bld.def(v1), op[0], op[1]); in emit_vop2_instruction() 902 bld.vop2(aco_opcode::v_mul_f32, Definition(dst), Operand::c32(0x3f800000u), tmp); in emit_vop2_instruction() 905 bld.nuw().vop2(opc, Definition(dst), op[0], op[1]); in emit_vop2_instruction() [all …]
|
D | aco_ir.h | 1259 VOP2_instruction& vop2() noexcept in vop2() function 1264 const VOP2_instruction& vop2() const noexcept in vop2() function
|
/third_party/mesa3d/docs/relnotes/ |
D | 21.0.0.rst | 2984 - aco: allow to use the range analysis UB in emit_{sop2,vop2}_instruction()
|