Skip to content

PMULHW

Jump to bottom
Henk-Jan Lebbink edited this page Jun 6, 2018 · 14 revisions

PMULHW — Multiply Packed Signed Integers and Store High Result

Opcode/ Instruction Op/ En 64/32 bit Mode Support CPUID Feature Flag Description
NP 0F E5 /r1 PMULHW mm, mm/m64 A V/V MMX Multiply the packed signed word integers in mm1 register and mm2/m64, and store the high 16 bits of the results in mm1.
66 0F E5 /r PMULHW xmm1, xmm2/m128 A V/V SSE2 Multiply the packed signed word integers in xmm1 and xmm2/m128, and store the high 16 bits of the results in xmm1.
VEX.NDS.128.66.0F.WIG E5 /r VPMULHW xmm1, xmm2, xmm3/m128 B V/V AVX Multiply the packed signed word integers in xmm2 and xmm3/m128, and store the high 16 bits of the results in xmm1.
VEX.NDS.256.66.0F.WIG E5 /r VPMULHW ymm1, ymm2, ymm3/m256 B V/V AVX2 Multiply the packed signed word integers in ymm2 and ymm3/m256, and store the high 16 bits of the results in ymm1.
EVEX.NDS.128.66.0F.WIG E5 /r VPMULHW xmm1 {k1}{z}, xmm2, xmm3/m128 C V/V AVX512VL AVX512BW Multiply the packed signed word integers in xmm2 and xmm3/m128, and store the high 16 bits of the results in xmm1 under writemask k1.
EVEX.NDS.256.66.0F.WIG E5 /r VPMULHW ymm1 {k1}{z}, ymm2, ymm3/m256 C V/V AVX512VL AVX512BW Multiply the packed signed word integers in ymm2 and ymm3/m256, and store the high 16 bits of the results in ymm1 under writemask k1.
EVEX.NDS.512.66.0F.WIG E5 /r VPMULHW zmm1 {k1}{z}, zmm2, zmm3/m512 C V/V AVX512BW Multiply the packed signed word integers in zmm2 and zmm3/m512, and store the high 16 bits of the results in zmm1 under writemask k1.
  1. See note in Section 2.4, “AVX and SSE Instruction Exception Specification” in the Intel® 64 and IA-32 Architectures Software Developer’s Manual, Volume 2A and Section 22.25.3, “Exception Conditions of Legacy SIMD Instructions Operating on MMX Registers” in the Intel® 64 and IA-32 Architectures Software Developer’s Manual, Volume 3A.

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 Mem ModRM:reg (w) EVEX.vvvv (r) ModRM:r/m (r) NA

Description

Performs a SIMD signed multiply of the packed signed word integers in the destination operand (first operand) and the source operand (second operand), and stores the high 16 bits of each intermediate 32-bit result in the destination operand. (Figure 4-12 shows this operation when using 64-bit operands.)

n 64-bit mode and not encoded with VEX/EVEX, using a REX prefix in the form of REX.R permits this instruction to access additional registers (XMM8-XMM15).

Legacy SSE version 64-bit operand: The source operand can be an MMX technology register or a 64-bit memory location. The destination operand is an MMX technology register.

128-bit Legacy SSE version: The first source and destination operands are XMM registers. The second source operand is an XMM register or a 128-bit memory location. Bits (MAXVL-1:128) of the corresponding YMM destination register remain unchanged.

VEX.128 encoded version: The first source and destination operands are XMM registers. The second source operand is an XMM register or a 128-bit memory location. Bits (MAXVL-1:128) of the destination YMM register are zeroed. VEX.L must be 0, otherwise the instruction will #UD. VEX.256 encoded version: The second source operand can be an YMM register or a 256-bit memory location. The first source and destination operands are YMM registers.

EVEX encoded versions: The first source operand is a ZMM/YMM/XMM register. The second source operand can be a ZMM/YMM/XMM register, a 512/256/128-bit memory location. The destination operand is a ZMM/YMM/XMM register conditionally updated with writemask k1.

Operation

PMULHW (with 64-bit operands)

    TEMP0[31:0] ←  DEST[15:0] ∗ SRC[15:0]; (* Signed multiplication *)
    TEMP1[31:0] ←  DEST[31:16] ∗ SRC[31:16];
    TEMP2[31:0] ←  DEST[47:32] ∗ SRC[47:32];
    TEMP3[31:0] ← DEST[63:48] ∗ SRC[63:48];
    DEST[15:0] ← 
                    TEMP0[31:16];
    DEST[31:16] ←  TEMP1[31:16];
    DEST[47:32] ←  TEMP2[31:16];
    DEST[63:48] ←  TEMP3[31:16];

PMULHW (with 128-bit operands)

    TEMP0[31:0] ←  DEST[15:0] ∗ SRC[15:0]; (* Signed multiplication *)
    TEMP1[31:0] ← DEST[31:16] ∗ SRC[31:16];
    TEMP2[31:0] ←  DEST[47:32] ∗ SRC[47:32];
    TEMP3[31:0] ← DEST[63:48] ∗ SRC[63:48];
    TEMP4[31:0] ← DEST[79:64] ∗ SRC[79:64];
    TEMP5[31:0] ←  DEST[95:80] ∗ SRC[95:80];
    TEMP6[31:0] ←  DEST[111:96] ∗ SRC[111:96];
    TEMP7[31:0] ← DEST[127:112] ∗ SRC[127:112];
    DEST[15:0] ← 
                    TEMP0[31:16];
    DEST[31:16] ←  TEMP1[31:16];
    DEST[47:32] ←  TEMP2[31:16];
    DEST[63:48] ←  TEMP3[31:16];
    DEST[79:64] ←  TEMP4[31:16];
    DEST[95:80] ←  TEMP5[31:16];
    DEST[111:96] ←  TEMP6[31:16];
    DEST[127:112] ← TEMP7[31:16];

VPMULHW (VEX.128 encoded version)

TEMP0[31:0] ← SRC1[15:0] * SRC2[15:0] (*Signed Multiplication*)
TEMP1[31:0] ← SRC1[31:16] * SRC2[31:16]
TEMP2[31:0] ← SRC1[47:32] * SRC2[47:32]
TEMP3[31:0] ← SRC1[63:48] * SRC2[63:48]
TEMP4[31:0] ← SRC1[79:64] * SRC2[79:64]
TEMP5[31:0] ← SRC1[95:80] * SRC2[95:80]
TEMP6[31:0] ← SRC1[111:96] * SRC2[111:96]
TEMP7[31:0] ← SRC1[127:112] * SRC2[127:112]
DEST[15:0] ← TEMP0[31:16]
DEST[31:16] ← TEMP1[31:16]
DEST[47:32] ← TEMP2[31:16]
DEST[63:48] ← TEMP3[31:16]
DEST[79:64] ← TEMP4[31:16]
DEST[95:80] ← TEMP5[31:16]
DEST[111:96] ← TEMP6[31:16]
DEST[127:112] ← TEMP7[31:16]
DEST[MAXVL-1:128] ← 0

PMULHW (VEX.256 encoded version)

TEMP0[31:0] ← SRC1[15:0] * SRC2[15:0] (*Signed Multiplication*)
TEMP1[31:0] ← SRC1[31:16] * SRC2[31:16]
TEMP2[31:0] ← SRC1[47:32] * SRC2[47:32]
TEMP3[31:0] ← SRC1[63:48] * SRC2[63:48]
TEMP4[31:0] ← SRC1[79:64] * SRC2[79:64]
TEMP5[31:0] ← SRC1[95:80] * SRC2[95:80]
TEMP6[31:0] ← SRC1[111:96] * SRC2[111:96]
TEMP7[31:0] ← SRC1[127:112] * SRC2[127:112]
TEMP8[31:0] ← SRC1[143:128] * SRC2[143:128]
TEMP9[31:0] ← SRC1[159:144] * SRC2[159:144]
TEMP10[31:0] ← SRC1[175:160] * SRC2[175:160]
TEMP11[31:0] ← SRC1[191:176] * SRC2[191:176]
TEMP12[31:0] ← SRC1[207:192] * SRC2[207:192]
TEMP13[31:0] ← SRC1[223:208] * SRC2[223:208]
TEMP14[31:0] ← SRC1[239:224] * SRC2[239:224]
TEMP15[31:0] ← SRC1[255:240] * SRC2[255:240]
DEST[15:0] ← TEMP0[31:16]
DEST[31:16] ← TEMP1[31:16]
DEST[47:32] ← TEMP2[31:16]
DEST[63:48] ← TEMP3[31:16]
DEST[79:64] ← TEMP4[31:16]
DEST[95:80] ← TEMP5[31:16]
DEST[111:96] ← TEMP6[31:16]
DEST[127:112] ← TEMP7[31:16]
DEST[143:128] ← TEMP8[31:16]
DEST[159:144] ← TEMP9[31:16]
DEST[175:160] ← TEMP10[31:16]
DEST[191:176] ← TEMP11[31:16]
DEST[207:192] ← TEMP12[31:16]
DEST[223:208] ← TEMP13[31:16]
DEST[239:224] ← TEMP14[31:16]
DEST[255:240] ← TEMP15[31:16]
DEST[MAXVL-1:256] ← 0

PMULHW (EVEX encoded versions)

(KL, VL) = (8, 128), (16, 256), (32, 512)
FOR j ← 0 TO KL-1
    i ← j * 16
    IF k1[j] OR *no writemask*
        THEN 
            temp[31:0] ← SRC1[i+15:i] * SRC2[i+15:i]
            DEST[i+15:i] ← tmp[31:16]
        ELSE 
            IF *merging-masking*
                            ; merging-masking
                THEN *DEST[i+15:i] remains unchanged*
                ELSE *zeroing-masking*
                            ; zeroing-masking
                    DEST[i+15:i] ← 0
            FI
    FI;
ENDFOR
DEST[MAXVL-1:VL] ← 0

Intel C/C++ Compiler Intrinsic Equivalent

VPMULHW __m512i _mm512_mulhi_epi16(__m512i a, __m512i b);
VPMULHW __m512i _mm512_mask_mulhi_epi16(__m512i s, __mmask32 k, __m512i a, __m512i b);
VPMULHW __m512i _mm512_maskz_mulhi_epi16( __mmask32 k, __m512i a, __m512i b);
VPMULHW __m256i _mm256_mask_mulhi_epi16(__m256i s, __mmask16 k, __m256i a, __m256i b);
VPMULHW __m256i _mm256_maskz_mulhi_epi16( __mmask16 k, __m256i a, __m256i b);
VPMULHW __m128i _mm_mask_mulhi_epi16(__m128i s, __mmask8 k, __m128i a, __m128i b);
VPMULHW __m128i _mm_maskz_mulhi_epi16( __mmask8 k, __m128i a, __m128i b);
PMULHW:__m64 _mm_mulhi_pi16 (__m64 m1, __m64 m2)
(V)PMULHW:__m128i _mm_mulhi_epi16 ( __m128i a, __m128i b)
VPMULHW:__m256i _mm256_mulhi_epi16 ( __m256i a, __m256i b)

Flags Affected

None.

SIMD Floating-Point Exceptions

None.

Other Exceptions

Non-EVEX-encoded instruction, see Exceptions Type 4.

EVEX-encoded instruction, see Exceptions Type E4.nb.

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

Clone this wiki locally