HADDPS—Packed Single-FP Horizontal Add

INSTRUCTION SET REFERENCE, A-L

3-430 Vol. 2A

See Figure 3-18 for HADDPS; see Figure 3-19 for VHADDPS.

Figure 3-19.  VHADDPS 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 (VLMAX-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 (VLMAX-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.

Figure 3-18.  HADDPS—Packed Single-FP Horizontal Add

OM15994

HADDPS xmm1, xmm2/m128

RESULT:
xmm1

xmm2/
m128

xmm1[31:0] + 

xmm1[63:32]

[31:0]

xmm1[95:64] + 

xmm1[127:96]

[63:32]

[63:32]

[31:0]

xmm1

[31:0]

[63:32]

xmm2/m128

[31:0] + xmm2/

m128[63:32]

[95:64]

xmm2/m128

[95:64] + xmm2/

m128[127:96]

[127:96]

[127:96]

[95:64]

[95:64]

[127:96]

Y6+Y7

X6+X7

Y2+Y3

X2+X3

DEST

SRC1

X0

SRC2

X1

X2

X3

X4

X5

X6

X7

Y0

Y1

Y2

Y3

Y4

Y5

Y6

Y7

X0+X1

Y4+Y5

X4+X5

Y0+Y1