Lines Matching defs:a16
1095 //>> v1: %a, v2b: %a16 = p_startpgm
1100 Temp a16 = inputs[1];
1105 //! v2b: %res0 = v_mul_f16 %res0_tmp, %a16
1107 writeout(0, fmul(u2u16(fneg(a)), a16));
1109 //! v2b: %res1_tmp = v_mul_f16 -1.0, %a16
1112 writeout(1, fmul(bld.as_uniform(fneg(a16)), a));
1114 //! v1: %res2_tmp = v_mul_f32 -1.0, %a16
1115 //! v2b: %res2 = v_mul_f16 %res2_tmp, %a16
1117 writeout(2, fmul(u2u16(bld.vop2_e64(aco_opcode::v_mul_f32, bld.def(v1), Operand::c32(0xbf800000u), bld.as_uniform(a16))), a16));
1124 //! v2b: %res4_tmp = v_mul_f16 %a16, %a16
1127 writeout(4, fsat(bld.as_uniform(fmul(a16, a16))));
1134 //! v2b: %res6_tmp = v_mul_f16 %a16, %a16
1137 writeout(6, fmul(bld.as_uniform(fmul(a16, a16)), bld.copy(bld.def(v1), Operand::c32(0x40000000))));
1140 //! v2b: %res7 = v_add_f16 %res7_tmp, %a16
1142 writeout(7, fadd(u2u16(fmul(a, a)), a16));
1144 //! v2b: %res8_tmp = v_mul_f16 %a16, %a16
1147 writeout(8, fadd(bld.as_uniform(fmul(a16, a16)), a));
1154 //! v2b: %res10_tmp = v_mul_f16 %a16, %a16
1157 writeout(10, bld.vop2_e64(aco_opcode::v_mul_f32, bld.def(v1), Operand::c32(0xbf800000u), bld.as_uniform(fmul(a16, a16))));
1164 //>> v1: %a, v2b: %a16 = p_startpgm
1171 Temp a16 = inputs[1];
1173 //! v1: %res0 = v_fma_mix_f32 %a, lo(%a16), -0
1175 writeout(0, fmul(a, f2f32(a16)));
1177 //! v1: %res1 = v_fma_mix_f32 1.0, %a, lo(%a16)
1179 writeout(1, fadd(a, f2f32(a16)));
1181 //! v1: %res2 = v_fma_mix_f32 1.0, lo(%a16), %a
1183 writeout(2, fadd(f2f32(a16), a));
1185 //! v1: %res3 = v_fma_mix_f32 %a, %a, lo(%a16)
1187 writeout(3, fma(a, a, f2f32(a16)));
1189 //! v1: %res4 = v_fma_mix_f32 %a, %a, lo(%a16)
1191 writeout(4, fma(a, a, f2f32(a16)));
1199 //>> v1: %a, v2b: %a16 = p_startpgm
1206 Temp a16 = inputs[1];
1209 //~gfx9! v1: %res0_cvt = v_cvt_f32_f16 %a16
1211 //~gfx10! v1: (precise)%res0 = v_fma_mix_f32 %a, %a, lo(%a16)
1213 writeout(0, fma(a, a, f2f32(a16), bld.precise()));
1216 //! v2b: (precise)%res1 = v_mul_f16 %a16, %res1_cvt
1218 writeout(1, fmul(a16, f2f16(a), bld.precise()));
1221 //! v2b: (precise)%res2 = v_add_f16 %a16, %res2_cvt
1223 writeout(2, fadd(a16, f2f16(a), bld.precise()));
1226 //! v2b: (precise)%res3 = v_fma_f16 %a16, %a16, %res3_cvt
1228 writeout(3, fma(a16, a16, f2f16(a), bld.precise()));
1232 //! v2b: %res4 = v_mul_f16 %a16, %res4_cvt
1234 writeout(4, fmul(a16, f2f16(a, bld.precise())));
1237 //! v2b: %res5 = v_add_f16 %a16, %res5_cvt
1239 writeout(5, fadd(a16, f2f16(a, bld.precise())));
1242 //! v2b: %res6 = v_fma_f16 %a16, %a16, %res6_cvt
1244 writeout(6, fma(a16, a16, f2f16(a, bld.precise())));
1252 //>> v1: %a, v2b: %a16 = p_startpgm
1259 Temp a16 = inputs[1];
1262 //! v1: %res0 = v_fma_mix_f32 -%a, lo(%a16), -0
1264 writeout(0, fmul(fneg(a), f2f32(a16)));
1266 //! v1: %res1 = v_fma_mix_f32 |%a|, lo(%a16), -0
1268 writeout(1, fmul(fabs(a), f2f32(a16)));
1271 //! v1: %res2 = v_fma_mix_f32 %a, -lo(%a16), -0
1273 writeout(2, fmul(a, fneg(f2f32(a16))));
1275 //! v1: %res3 = v_fma_mix_f32 %a, -lo(%a16), -0
1277 writeout(3, fmul(a, f2f32(fneg(a16))));
1280 //! v1: %res4 = v_fma_mix_f32 %a, |lo(%a16)|, -0
1282 writeout(4, fmul(a, fabs(f2f32(a16))));
1284 //! v1: %res5 = v_fma_mix_f32 %a, |lo(%a16)|, -0
1286 writeout(5, fmul(a, f2f32(fabs(a16))));
1289 //! v1: %res6 = v_fma_mix_f32 %a, -|lo(%a16)|, -0
1291 writeout(6, fmul(a, fneg(f2f32(fabs(a16)))));
1293 //! v1: %res7 = v_fma_mix_f32 %a, |lo(%a16)|, -0
1295 writeout(7, fmul(a, fabs(f2f32(fabs(a16)))));
1297 //! v1: %res8 = v_fma_mix_f32 %a, -|lo(%a16)|, -0
1299 writeout(8, fmul(a, fneg(fabs(f2f32(fabs(a16))))));
1301 //! v1: %res9 = v_fma_mix_f32 %a, -|lo(%a16)|, -0
1303 writeout(9, fmul(a, f2f32(fneg(fabs(a16)))));
1305 //! v1: %res10 = v_fma_mix_f32 %a, |lo(%a16)|, -0
1307 writeout(10, fmul(a, fneg(f2f32(fneg(fabs(a16))))));
1309 //! v1: %res11 = v_fma_mix_f32 %a, |lo(%a16)|, -0
1311 writeout(11, fmul(a, fabs(f2f32(fneg(fabs(a16))))));
1313 //! v1: %res12 = v_fma_mix_f32 %a, -|lo(%a16)|, -0
1315 writeout(12, fmul(a, fneg(fabs(f2f32(fneg(fabs(a16)))))));
1347 //>> v1: %a, v1: %b, v1: %c, v2b: %a16, v2b: %b16 = p_startpgm
1356 Temp a16 = inputs[3];
1371 //! v2b: %res3 = v_fma_mixlo_f16 lo(%a16), %b, -0
1373 writeout(3, f2f16(fmul(f2f32(a16), b)));
1389 //>> v2b: %a16 = p_startpgm
1395 Temp a16 = inputs[0];
1397 //! v2b: %res0_tmp = v_mul_f16 %a16, %a16
1400 writeout(0, f2f32(fmul(a16, a16), bld.precise()));
1402 //! v2b: (precise)%res1_tmp = v_mul_f16 %a16, %a16
1405 writeout(1, f2f32(fmul(a16, a16, bld.precise())));
1413 //>> v1: %a, v1: %b, v2b: %a16, v2b: %b16 = p_startpgm
1421 Temp a16 = inputs[2];
1438 writeout(2, f2f32(fabs(fadd(a16, b16))));
1443 writeout(3, f2f32(fneg(fadd(a16, b16))));
1462 //>> v1: %a, v1: %b, v1: %c, v2b: %a16, v2b: %c16 = p_startpgm
1471 Temp a16 = inputs[3];
1474 //! v1: %res0 = v_fma_mix_f32 lo(%a16), %b, %c
1476 writeout(0, fadd(fmul(f2f32(a16), b), c));
1483 //! v1: %res2_mul = v_fma_mix_f32 lo(%a16), %b, -0
1486 writeout(2, bld.vop2(aco_opcode::v_mul_f32, bld.def(v1), Operand::c32(0x40000000), fadd(fmul(f2f32(a16), b), c)));
1489 //! v1: %res3 = v_fma_mix_f32 -lo(%a16), %b, |lo(%c16)|
1491 writeout(3, fadd(fmul(fneg(f2f32(a16)), b), fabs(f2f32(c16))));
1516 //>> v1: %a, v1: %b, v1: %c, v2b: %a16, v2b: %b16 = p_startpgm
1525 Temp a16 = inputs[3];
1529 //~gfx9! v1: %res0 = v_fma_mix_f32 lo(%a16), %b, %c
1530 //~gfx10! v1: (precise)%res0_tmp = v_fma_mix_f32 lo(%a16), %b, -0
1533 writeout(0, fadd(fmul(f2f32(a16), b, bld.precise()), c));
1535 //~gfx9! v1: (precise)%res1 = v_fma_mix_f32 lo(%a16), %b, %c
1536 //~gfx10! v1: %res1_tmp = v_fma_mix_f32 lo(%a16), %b, -0
1539 writeout(1, fadd(fmul(f2f32(a16), b), c, bld.precise()));
1552 //! v2b: %res4_tmp = v_mul_f16 %a16, %b16
1555 writeout(4, f2f32(fmul(a16, b16)));
1557 //! v2b: %res5_tmp = v_add_f16 %a16, %b16
1560 writeout(5, f2f32(fadd(a16, b16)));
1563 //! v2b: %res6 = v_add_f16 %res6_tmp, %a16
1565 writeout(6, fadd(f2f16(fmul(a, b)), a16));
1567 //! v2b: %res7_tmp = v_mul_f16 %a16, %b16
1570 writeout(7, fadd(f2f32(fmul(a16, b16)), c));
1578 //>> v1: %a, v2b: %a16 = p_startpgm
1585 Temp a16 = inputs[1];
1587 //! v1: %res0 = v_fma_mix_f32 lo(%a16), %a, -0 clamp
1589 writeout(0, fsat(fmul(f2f32(a16), a)));
1605 //>> v1: %a, v2b: %a16 = p_startpgm
1612 Temp a16 = inputs[1];
1618 //! v1: %res0_cvt = v_cvt_f32_f16 %a16
1619 //! v2b: %res0 = v_mul_f16 %res0_cvt, %a16
1621 writeout(0, fmul(u2u16(f2f32(a16)), a16));
1628 //! v2b: %res2_mul = v_mul_f16 %a16, %a16
1631 writeout(2, f2f16(bld.as_uniform(fmul(a16, a16))));
1638 //! v1: %res4_mul = v_fma_mix_f32 lo(%a16), %a, -0
1641 writeout(4, fsat(u2u16(fmul(f2f32(a16), a))));
1653 //! v2b: %res7_mul = v_mul_f16 %a16, %a16
1654 //! v1: %res7 = v_fma_mix_f32 1.0, %res7_mul, lo(%a16)
1656 writeout(7, fadd(bld.as_uniform(fmul(a16, a16)), f2f32(a16)));
1659 //! v1: %res8 = v_fma_mix_f32 lo(%a16), %a16, -0
1661 writeout(8, fmul(f2f32(a16), a16));