YMM 509 test/correctness/simd_op_check.cpp check("vsqrtps" YMM, 8, sqrt(f32_1)); YMM 510 test/correctness/simd_op_check.cpp check("vsqrtpd" YMM, 4, sqrt(f64_1)); YMM 511 test/correctness/simd_op_check.cpp check(use_avx512_skylake ? "vrsqrt14ps" : "vrsqrtps" YMM, 8, fast_inverse_sqrt(f32_1)); YMM 512 test/correctness/simd_op_check.cpp check(use_avx512_skylake ? "vrcp14ps" : "vrcpps" YMM, 8, fast_inverse(f32_1)); YMM 522 test/correctness/simd_op_check.cpp check("vaddps" YMM, 8, f32_1 + f32_2); YMM 523 test/correctness/simd_op_check.cpp check("vaddpd" YMM, 4, f64_1 + f64_2); YMM 524 test/correctness/simd_op_check.cpp check("vmulps" YMM, 8, f32_1 * f32_2); YMM 525 test/correctness/simd_op_check.cpp check("vmulpd" YMM, 4, f64_1 * f64_2); YMM 526 test/correctness/simd_op_check.cpp check("vsubps" YMM, 8, f32_1 - f32_2); YMM 527 test/correctness/simd_op_check.cpp check("vsubpd" YMM, 4, f64_1 - f64_2); YMM 531 test/correctness/simd_op_check.cpp check("vminps" YMM, 8, min(f32_1, f32_2)); YMM 532 test/correctness/simd_op_check.cpp check("vminpd" YMM, 4, min(f64_1, f64_2)); YMM 533 test/correctness/simd_op_check.cpp check("vmaxps" YMM, 8, max(f32_1, f32_2)); YMM 534 test/correctness/simd_op_check.cpp check("vmaxpd" YMM, 4, max(f64_1, f64_2)); YMM 535 test/correctness/simd_op_check.cpp check("vroundps" YMM, 8, round(f32_1)); YMM 536 test/correctness/simd_op_check.cpp check("vroundpd" YMM, 4, round(f64_1)); YMM 538 test/correctness/simd_op_check.cpp check("vcmpeqpd" YMM, 4, select(f64_1 == f64_2, 1.0f, 2.0f)); YMM 541 test/correctness/simd_op_check.cpp check("vcmpltpd" YMM, 4, select(f64_1 < f64_2, 1.0f, 2.0f)); YMM 542 test/correctness/simd_op_check.cpp check("vcmpeqps" YMM, 8, select(f32_1 == f32_2, 1.0f, 2.0f)); YMM 545 test/correctness/simd_op_check.cpp check("vcmpltps" YMM, 8, select(f32_1 < f32_2, 1.0f, 2.0f)); YMM 548 test/correctness/simd_op_check.cpp check(use_avx512_skylake ? "vinsertf32x8" : "vblend*ps" YMM, 8, select(f32_1 > 0.7f, f32_1, f32_2)); YMM 549 test/correctness/simd_op_check.cpp check(use_avx512 ? "vinsertf64x4" : "vblend*pd" YMM, 4, select(f64_1 > cast<double>(0.7f), f64_1, f64_2)); YMM 551 test/correctness/simd_op_check.cpp check("vcvttps2dq" YMM, 8, i32(f32_1)); YMM 552 test/correctness/simd_op_check.cpp check("vcvtdq2ps" YMM, 8, f32(i32_1)); YMM 554 test/correctness/simd_op_check.cpp check("vcvtdq2pd" YMM, 8, f64(i32_1)); YMM 555 test/correctness/simd_op_check.cpp check("vcvtps2pd" YMM, 8, f64(f32_1)); YMM 565 test/correctness/simd_op_check.cpp check("vpaddb" YMM, 32, u8_1 + u8_2); YMM 566 test/correctness/simd_op_check.cpp check("vpsubb" YMM, 32, u8_1 - u8_2); YMM 571 test/correctness/simd_op_check.cpp check("vpaddw" YMM, 16, u16_1 + u16_2); YMM 572 test/correctness/simd_op_check.cpp check("vpsubw" YMM, 16, u16_1 - u16_2); YMM 577 test/correctness/simd_op_check.cpp check("vpaddd" YMM, 8, i32_1 + i32_2); YMM 578 test/correctness/simd_op_check.cpp check("vpsubd" YMM, 8, i32_1 - i32_2); YMM 579 test/correctness/simd_op_check.cpp check("vpmulhw" YMM, 16, i16((i32(i16_1) * i32(i16_2)) / (256*256))); YMM 580 test/correctness/simd_op_check.cpp check("vpmulhw" YMM, 16, i16((i32(i16_1) * i32(i16_2)) >> 16)); YMM 581 test/correctness/simd_op_check.cpp check("vpmullw" YMM, 16, i16_1 * i16_2); YMM 583 test/correctness/simd_op_check.cpp check("vpcmp*b" YMM, 32, select(u8_1 == u8_2, u8(1), u8(2))); YMM 584 test/correctness/simd_op_check.cpp check("vpcmp*b" YMM, 32, select(u8_1 > u8_2, u8(1), u8(2))); YMM 585 test/correctness/simd_op_check.cpp check("vpcmp*w" YMM, 16, select(u16_1 == u16_2, u16(1), u16(2))); YMM 586 test/correctness/simd_op_check.cpp check("vpcmp*w" YMM, 16, select(u16_1 > u16_2, u16(1), u16(2))); YMM 587 test/correctness/simd_op_check.cpp check("vpcmp*d" YMM, 8, select(u32_1 == u32_2, u32(1), u32(2))); YMM 588 test/correctness/simd_op_check.cpp check("vpcmp*d" YMM, 8, select(u32_1 > u32_2, u32(1), u32(2))); YMM 592 test/correctness/simd_op_check.cpp check("vpmaxsw" YMM, 16, max(i16_1, i16_2)); YMM 593 test/correctness/simd_op_check.cpp check("vpminsw" YMM, 16, min(i16_1, i16_2)); YMM 594 test/correctness/simd_op_check.cpp check("vpmaxub" YMM, 32, max(u8_1, u8_2)); YMM 595 test/correctness/simd_op_check.cpp check("vpminub" YMM, 32, min(u8_1, u8_2)); YMM 596 test/correctness/simd_op_check.cpp check("vpmulhuw" YMM, 16, u16((u32(u16_1) * u32(u16_2))/(256*256))); YMM 597 test/correctness/simd_op_check.cpp check("vpmulhuw" YMM, 16, u16((u32(u16_1) * u32(u16_2))>>16)); YMM 599 test/correctness/simd_op_check.cpp check("vpaddq" YMM, 8, i64_1 + i64_2); YMM 600 test/correctness/simd_op_check.cpp check("vpsubq" YMM, 8, i64_1 - i64_2); YMM 615 test/correctness/simd_op_check.cpp check("vpmuludq" YMM, 8, u64(u32_1) * u64(u32_2)); YMM 617 test/correctness/simd_op_check.cpp check("vpmulld" YMM, 8, i32_1 * i32_2); YMM 619 test/correctness/simd_op_check.cpp check("vpblend*b" YMM, 32, select(u8_1 > 7, u8_1, u8_2)); YMM 621 test/correctness/simd_op_check.cpp check("vpmaxsb" YMM, 32, max(i8_1, i8_2)); YMM 622 test/correctness/simd_op_check.cpp check("vpminsb" YMM, 32, min(i8_1, i8_2)); YMM 623 test/correctness/simd_op_check.cpp check("vpmaxuw" YMM, 16, max(u16_1, u16_2)); YMM 624 test/correctness/simd_op_check.cpp check("vpminuw" YMM, 16, min(u16_1, u16_2)); YMM 625 test/correctness/simd_op_check.cpp check("vpmaxud" YMM, 16, max(u32_1, u32_2)); YMM 626 test/correctness/simd_op_check.cpp check("vpminud" YMM, 16, min(u32_1, u32_2)); YMM 627 test/correctness/simd_op_check.cpp check("vpmaxsd" YMM, 8, max(i32_1, i32_2)); YMM 628 test/correctness/simd_op_check.cpp check("vpminsd" YMM, 8, min(i32_1, i32_2)); YMM 630 test/correctness/simd_op_check.cpp check("vpcmpeqq" YMM, 4, select(i64_1 == i64_2, i64(1), i64(2))); YMM 632 test/correctness/simd_op_check.cpp check("vpcmpgtq" YMM, 4, select(i64_1 > i64_2, i64(1), i64(2)));