MOVSHDUP

MOVSHDUP copies the odd singles of its source over the even ones: the result is elements 1, 1, 3 and 3. MOVSLDUP repeats the even ones instead, and MOVDDUP a double; all three came with SSE3, beyond the first x86-64 level, and VMOVSHDUP does the same over every pair of a wider register. Unlike MOVDDUP, which reads eight bytes from anywhere, the memory form reads all 16 bytes and needs them aligned to 16. The result keeps bits 128 and up of the register (see SSE and AVX), and a NaN or denormal is copied as it is.

Zen 5’s recorded movshdup xmm12, xmmword ptr [mem] gave bytes 4 to 7 of memory in elements 0 and 1 and bytes 12 to 15 in elements 2 and 3, and the register form did the same with XMM6.

Uses#

• Element 1 to element 0, without a copy. It writes a separate register, so it replaces the MOVAPS and SHUFPS pair that otherwise brings element 1 down. In our builds at -O2 with -march=x86-64-v2, Clang 23 starts a sum of the four floats of a vector with MOVSHDUP and ADDSS, and takes the imaginary part of a float complex returned in XMM0 with movshdup xmm5, xmm0.
• Complex products. With -fcx-limited-range, GCC 14 multiplies two float complex values with MOVSLDUP and MOVSHDUP, which spread the real and the imaginary part of the first, SHUFPS with E1h, which swaps the parts of the second, two MULPS and an ADDSUBPS, the single-precision pattern of ADDSUBPD.
• As a shuffle. Both compilers emit it for __builtin_shufflevector(v, v, 1, 1, 3, 3) with SSE3 available; for the default target the same shuffle is shufps xmm0, xmm0, 245.

Forms

Data flow

Where the elements of the sources go, element 0 on the right; results that depend on a control value are drawn for an example. How to read.

Operation

What the forms do, in pseudocode that our tests run against results recorded on real processors. How to read.

for i in range(4):
    j = i - i % 2 + 1
    xmm1.d[i] = src2.d[j]
# bits 128 and up of the register are unchanged

Operands

Flags

None: XED records no flag that the forms test or change.

Exceptions

The exception class is XED's name for the set of exception conditions that a form follows. How to read.

All forms

SSE_TYPE_4

Attributes

What else XED records about the forms. How to read.

Memory forms

aligned memory operand

Intrinsics

The C intrinsics of Clang 23.1.2 that compile to MOVSHDUP, each with the form it compiles to and the Clang options it needs. The forms come from compiling every intrinsic and decoding the result with Intel XED; all intrinsics A–Z explains the method.

CPUID and processors

Each form belongs to an XED ISA set. A form can be used when the processor reports every CPUID bit of one of its ISA set's alternatives.

More SSE3 instructions

• ADDSUBPD
• ADDSUBPS
• FISTTP
• HADDPD
• HADDPS
• HSUBPD
• HSUBPS
• LDDQU
• MOVDDUP
• MOVSLDUP