Skip to content

UNPCKHPS

Jump to bottom
Henk-Jan Lebbink edited this page Jun 5, 2018 · 14 revisions

UNPCKHPS — Unpack and Interleave High Packed Single-Precision Floating-Point Values

Opcode/ Instruction Op / En 64/32 bit Mode Support CPUID Feature Flag Description
NP 0F 15 /r UNPCKHPS xmm1, xmm2/m128 A V/V SSE Unpacks and Interleaves single-precision floating-point values from high quadwords of xmm1 and xmm2/m128.
VEX.NDS.128.0F.WIG 15 /r VUNPCKHPS xmm1, xmm2, xmm3/m128 B V/V AVX Unpacks and Interleaves single-precision floating-point values from high quadwords of xmm2 and xmm3/m128.
VEX.NDS.256.0F.WIG 15 /r VUNPCKHPS ymm1, ymm2, ymm3/m256 B V/V AVX Unpacks and Interleaves single-precision floating-point values from high quadwords of ymm2 and ymm3/m256.
EVEX.NDS.128.0F.W0 15 /r VUNPCKHPS xmm1 {k1}{z}, xmm2, xmm3/m128/m32bcst C V/V AVX512VL AVX512F Unpacks and Interleaves single-precision floating-point values from high quadwords of xmm2 and xmm3/m128/m32bcst and write result to xmm1 subject to writemask k1.
EVEX.NDS.256.0F.W0 15 /r VUNPCKHPS ymm1 {k1}{z}, ymm2, ymm3/m256/m32bcst C V/V AVX512VL AVX512F Unpacks and Interleaves single-precision floating-point values from high quadwords of ymm2 and ymm3/m256/m32bcst and write result to ymm1 subject to writemask k1.
EVEX.NDS.512.0F.W0 15 /r VUNPCKHPS zmm1 {k1}{z}, zmm2, zmm3/m512/m32bcst C V/V AVX512F Unpacks and Interleaves single-precision floating-point values from high quadwords of zmm2 and zmm3/m512/m32bcst and write result to zmm1 subject to writemask k1.

Instruction Operand Encoding

Op/En Tuple Type Operand 1 Operand 2 Operand 3 Operand 4
A NA ModRM:reg (r, w) ModRM:r/m (r) NA NA
B NA ModRM:reg (w) VEX.vvvv (r) ModRM:r/m (r) NA
C Full ModRM:reg (w) EVEX.vvvv (r) ModRM:r/m (r) NA

Description

Performs an interleaved unpack of the high single-precision floating-point values from the first source operand and the second source operand.

128-bit Legacy SSE version: The second source can be an XMM register or an 128-bit memory location. The destination is not distinct from the first source XMM register and the upper bits (MAXVL-1:128) of the corresponding ZMM register destination are unmodified. When unpacking from a memory operand, an implementation may fetch only the appropriate 64 bits; however, alignment to 16-byte boundary and normal segment checking will still be enforced.

VEX.128 encoded version: The first source operand is a XMM register. The second source operand can be a XMM register or a 128-bit memory location. The destination operand is a XMM register. The upper bits (MAXVL-1:128) of the corresponding ZMM register destination are zeroed.

VEX.256 encoded version: The second source operand is an YMM register or an 256-bit memory location. The first source operand and destination operands are YMM registers.

SRC1 X7 SRC2 Y7 DEST Y7 X6 Y6 X7 X5 Y5 Y6 X4 Y4 X6 X3 Y3 Y3 X2 Y2 X3 X1 Y1 Y2 X0 Y0 X2
X7 X6 X5 X4 X3 X2 X1 X0
Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0
Y7 X7 Y6 X6 Y3 X3 Y2 X2

Figure 4-27. VUNPCKHPS Operation

EVEX.512 encoded version: The first source operand is a ZMM register. The second source operand is a ZMM register, a 512-bit memory location, or a 512-bit vector broadcasted from a 32-bit memory location. The destination operand is a ZMM register, conditionally updated using writemask k1.

EVEX.256 encoded version: The first source operand is a YMM register. The second source operand is a YMM register, a 256-bit memory location, or a 256-bit vector broadcasted from a 32-bit memory location. The destination operand is a YMM register, conditionally updated using writemask k1.

EVEX.128 encoded version: The first source operand is a XMM register. The second source operand is a XMM register, a 128-bit memory location, or a 128-bit vector broadcasted from a 32-bit memory location. The destination operand is a XMM register, conditionally updated using writemask k1.

Operation

VUNPCKHPS (EVEX encoded version when SRC2 is a register)

(KL, VL) = (4, 128), (8, 256), (16, 512)
IF VL >= 128
    TMP_DEST[31:0] ← SRC1[95:64]
    TMP_DEST[63:32] ← SRC2[95:64]
    TMP_DEST[95:64] ← SRC1[127:96]
    TMP_DEST[127:96] ← SRC2[127:96]
FI;
IF VL >= 256
    TMP_DEST[159:128] ← SRC1[223:192]
    TMP_DEST[191:160] ← SRC2[223:192]
    TMP_DEST[223:192] ← SRC1[255:224]
    TMP_DEST[255:224] ← SRC2[255:224]
FI;
IF VL >= 512
    TMP_DEST[287:256] ← SRC1[351:320]
    TMP_DEST[319:288] ← SRC2[351:320]
    TMP_DEST[351:320] ← SRC1[383:352]
    TMP_DEST[383:352] ← SRC2[383:352]
    TMP_DEST[415:384] ← SRC1[479:448]
    TMP_DEST[447:416] ← SRC2[479:448]
    TMP_DEST[479:448] ← SRC1[511:480]
    TMP_DEST[511:480] ← SRC2[511:480]
FI;
FOR j ← 0 TO KL-1
    i ← j * 32
    IF k1[j] OR *no writemask*
        THEN DEST[i+31:i] ← TMP_DEST[i+31:i]
        ELSE 
            IF *merging-masking*
                            ; merging-masking
                THEN *DEST[i+31:i] remains unchanged*
                ELSE *zeroing-masking*
                            ; zeroing-masking
                    DEST[i+31:i] ← 0
            FI
    FI;
ENDFOR
DEST[MAXVL-1:VL] ← 0

VUNPCKHPS (EVEX encoded version when SRC2 is memory)

(KL, VL) = (4, 128), (8, 256), (16, 512)
FOR j ← 0 TO KL-1
    i ← j * 32
    IF (EVEX.b = 1)
        THEN TMP_SRC2[i+31:i] ← SRC2[31:0]
        ELSE TMP_SRC2[i+31:i] ← SRC2[i+31:i]
    FI;
ENDFOR;
IF VL >= 128
    TMP_DEST[31:0] ← SRC1[95:64]
    TMP_DEST[63:32] ← TMP_SRC2[95:64]
    TMP_DEST[95:64] ← SRC1[127:96]
    TMP_DEST[127:96] ← TMP_SRC2[127:96]
FI;
IF VL >= 256
    TMP_DEST[159:128] ← SRC1[223:192]
    TMP_DEST[191:160] ← TMP_SRC2[223:192]
    TMP_DEST[223:192] ← SRC1[255:224]
    TMP_DEST[255:224] ← TMP_SRC2[255:224]
FI;
IF VL >= 512
    TMP_DEST[287:256] ← SRC1[351:320]
    TMP_DEST[319:288] ← TMP_SRC2[351:320]
    TMP_DEST[351:320] ← SRC1[383:352]
    TMP_DEST[383:352] ← TMP_SRC2[383:352]
    TMP_DEST[415:384] ← SRC1[479:448]
    TMP_DEST[447:416] ← TMP_SRC2[479:448]
    TMP_DEST[479:448] ← SRC1[511:480]
    TMP_DEST[511:480] ← TMP_SRC2[511:480]
FI;
FOR j ← 0 TO KL-1
    i ← j * 32
    IF k1[j] OR *no writemask*
        THEN DEST[i+31:i] ← TMP_DEST[i+31:i]
        ELSE 
            IF *merging-masking*
                            ; merging-masking
                THEN *DEST[i+31:i] remains unchanged*
                ELSE *zeroing-masking*
                            ; zeroing-masking
                    DEST[i+31:i] ← 0
            FI
    FI;
ENDFOR
DEST[MAXVL-1:VL] ← 0

VUNPCKHPS (VEX.256 encoded version)

DEST[31:0] ←SRC1[95:64]
DEST[63:32] ←SRC2[95:64]
DEST[95:64] ←SRC1[127:96]
DEST[127:96] ←SRC2[127:96]
DEST[159:128] ←SRC1[223:192]
DEST[191:160] ←SRC2[223:192]
DEST[223:192] ←SRC1[255:224]
DEST[255:224] ←SRC2[255:224]
DEST[MAXVL-1:256] ← 0

VUNPCKHPS (VEX.128 encoded version)

DEST[31:0] ←SRC1[95:64]
DEST[63:32] ←SRC2[95:64]
DEST[95:64] ←SRC1[127:96]
DEST[127:96] ←SRC2[127:96]
DEST[MAXVL-1:128] ←0

UNPCKHPS (128-bit Legacy SSE version)

DEST[31:0] ←SRC1[95:64]
DEST[63:32] ←SRC2[95:64]
DEST[95:64] ←SRC1[127:96]
DEST[127:96] ←SRC2[127:96]
DEST[MAXVL-1:128] (Unmodified)

Intel C/C++ Compiler Intrinsic Equivalent

VUNPCKHPS __m512 _mm512_unpackhi_ps( __m512 a, __m512 b);
VUNPCKHPS __m512 _mm512_mask_unpackhi_ps(__m512 s, __mmask16 k, __m512 a, __m512 b);
VUNPCKHPS __m512 _mm512_maskz_unpackhi_ps(__mmask16 k, __m512 a, __m512 b);
VUNPCKHPS __m256 _mm256_unpackhi_ps (__m256 a, __m256 b);
VUNPCKHPS __m256 _mm256_mask_unpackhi_ps(__m256 s, __mmask8 k, __m256 a, __m256 b);
VUNPCKHPS __m256 _mm256_maskz_unpackhi_ps(__mmask8 k, __m256 a, __m256 b);
UNPCKHPS __m128 _mm_unpackhi_ps (__m128 a, __m128 b);
VUNPCKHPS __m128 _mm_mask_unpackhi_ps(__m128 s, __mmask8 k, __m128 a, __m128 b);
VUNPCKHPS __m128 _mm_maskz_unpackhi_ps(__mmask8 k, __m128 a, __m128 b);

SIMD Floating-Point Exceptions

None

Other Exceptions

Non-EVEX-encoded instructions, see Exceptions Type 4. EVEX-encoded instructions, see Exceptions Type E4NF.

Source: Intel® Architecture Software Developer's Manual (May 2018)
Generated: 5-6-2018

Clone this wiki locally