HADDPD—Packed Double Precision Floating-Point Horizontal Add

Opcode/Instruction Op/En 64/32-bit Mode CPUID Feature Flag Description

66 0F 7C /r

HADDPD xmm1, xmm2/m128

RM V/V SSE3 Horizontal add packed double precision floating-point values from xmm2/m128 to xmm1.

VEX.128.66.0F.WIG 7C /r

VHADDPD xmm1,xmm2, xmm3/m128

RVM V/V AVX Horizontal add packed double precision floating-point values from xmm2 and xmm3/mem.

VEX.256.66.0F.WIG 7C /r

VHADDPD ymm1, ymm2, ymm3/m256

RVM V/V AVX Horizontal add packed double precision floating-point values from ymm2 and ymm3/mem.

Instruction Operand Encoding

Op/En Operand 1 Operand 2 Operand 3 Operand 4
RM ModRM:reg (r, w) ModRM:r/m (r) N/A N/A
RVM ModRM:reg (w) VEX.vvvv (r) ModRM:r/m (r) N/A

Description

Adds the double precision floating-point values in the high and low quadwords of the destination operand and stores the result in the low quadword of the destination operand.

Adds the double precision floating-point values in the high and low quadwords of the source operand and stores the result in the high quadword of the destination operand.

In 64-bit mode, use of the REX.R prefix permits this instruction to access additional registers (XMM8-XMM15).

See Figure 3-17 for HADDPD; see Figure 3-18 for VHADDPD.

HADDPD xmm1, xmm2/m128 xmm2 [127:64] [63:0] /m128 xmm1 [127:64] [63:0] Result: xmm2/m128[63:0] + xmm1[63:0] + xmm1[127:64] xmm1 xmm2/m128[127:64] [127:64] [63:0]

OM15993

Figure 3-17. HADDPD—Packed Double Precision Floating-Point Horizontal Add

X3

X2

X1

X0

Y3 Y2 Y1 Y0

SRC1

SRC2

DEST

Y2 + Y3

X2 + X3

Y0 + Y1

X0 + X1

Figure 3-18. VHADDPD Operation

128-bit Legacy SSE version: The second source can be an XMM register or an 128-bit memory location. The desti-nation is not distinct from the first source XMM register and the upper bits (MAXVL-1:128) of the corresponding YMM register destination are unmodified.

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

VEX.256 encoded version: The first source operand is a YMM register. The second source operand can be a YMM register or a 256-bit memory location. The destination operand is a YMM register.

Operation

HADDPD (128-bit Legacy SSE Version)

DEST[63:0] := SRC1[127:64] + SRC1[63:0]
DEST[127:64] := SRC2[127:64] + SRC2[63:0]
DEST[MAXVL-1:128] (Unmodified)

VHADDPD (VEX.128 Encoded Version)

DEST[63:0] := SRC1[127:64] + SRC1[63:0]
DEST[127:64] := SRC2[127:64] + SRC2[63:0]
DEST[MAXVL-1:128] := 0

VHADDPD (VEX.256 Encoded Version)

DEST[63:0] := SRC1[127:64] + SRC1[63:0]
DEST[127:64] := SRC2[127:64] + SRC2[63:0]
DEST[191:128] := SRC1[255:192] + SRC1[191:128]
DEST[255:192] := SRC2[255:192] + SRC2[191:128]

Intel C/C++ Compiler Intrinsic Equivalent

VHADDPD __m256d _mm256_hadd_pd (__m256d a, __m256d b);

HADDPD __m128d _mm_hadd_pd (__m128d a, __m128d b);

Exceptions

When the source operand is a memory operand, the operand must be aligned on a 16-byte boundary or a general-protection exception (#GP) will be generated.

Numeric Exceptions

Overflow, Underflow, Invalid, Precision, Denormal.

Other Exceptions

See Table 2-19, “Type 2 Class Exception Conditions.”