VPBROADCASTD

VPBROADCASTD fills every doubleword of a vector with one value: the low doubleword of an XMM register, a doubleword in memory, or, in the EVEX forms, a general-purpose register. Vector code uses it to set up a constant or a loop-invariant value in every element, which is what _mm256_set1_epi32(x) asks for.

The sources differ by encoding:

• VEX (AVX2) takes an XMM register or memory. A value in a general-purpose register first goes to an XMM register: GCC 14 and Clang 23 at -O2 compile _mm256_set1_epi32(x) to VMOVD and VPBROADCASTD. AVX itself could broadcast only from memory, with VBROADCASTSS, which copies the same 32 bits; for AVX2, Clang emits that instruction for a broadcast of an integer from memory.
• EVEX (AVX-512F) adds the general-purpose register source (opcode 7C) and masking, so with AVX-512 both compilers broadcast straight from the register.

The memory form reads only the four bytes of the value, at any alignment, so it never touches the bytes around it. That makes it a compact way to load a repeated constant: a 4-byte literal instead of a 32- or 64-byte one, which is how Clang loads such constants with AVX2. With AVX-512 Clang goes further and uses embedded broadcast, a memory operand marked {1to8} or {1to16} that the arithmetic instruction itself repeats across the vector, with no VPBROADCASTD at all. GCC 14 instead builds the constant in a general-purpose register and broadcasts it from there.

VPBROADCASTB, VPBROADCASTW and VPBROADCASTQ do the same for other element sizes, VBROADCASTI128 repeats 16 bytes, and VPBROADCASTMW2D copies a mask register into every element.

In the optimization guide

What the optimization guide says about VPBROADCASTD, a line per claim. Each links to the claim in the guide, with its sources and where they differ.

Skylake to Cooper Lake

• AVX-512 and frequency on the servers: A broadcast of a 32-bit or 64-bit element from memory runs on the load ports, sparing port 5; a 16-bit one still goes through port 5, so it gains nothing from the memory form.

Forms

Latency and throughput

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(VL // 32):
    if k1[i]:  # 1 without a write mask
        zmm1.d[i] = src2.d[0]
    elif zeroing:
        zmm1.d[i] = 0
    # else: zmm1.d[i] keeps its value
# zmm1 is zeroed from bit VL up

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.

VEX forms

avx-type-6

VPBROADCASTD xmm {k}{z}, xmm/m32 VPBROADCASTD ymm {k}{z}, xmm/m32 VPBROADCASTD zmm {k}{z}, xmm/m32

AVX512-E6

VPBROADCASTD xmm {k}{z}, r32 VPBROADCASTD ymm {k}{z}, r32 VPBROADCASTD zmm {k}{z}, r32

AVX512-E7NM

Attributes

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

EVEX forms

write masking

EVEX memory forms

memory fault suppression

Intrinsics

The C intrinsics of Clang 23.1.2 that compile to VPBROADCASTD, 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 AVX2 instructions

• VPACKSSWB
• VPACKUSDW
• VPACKUSWB
• VPADDB
• VPADDD
• VPADDQ
• VPADDSB
• VPADDSW
• VPADDUSB
• VPADDUSW
• VPADDW
• VPALIGNR
• VPAND
• VPANDN
• VPAVGB
• VPAVGW
• VPBLENDD
• VPBLENDVB
• VPBLENDW
• VPBROADCASTB
• VPBROADCASTQ
• VPBROADCASTW
• VPCMPEQB
• VPCMPEQD
• VPCMPEQQ
• VPCMPEQW
• VPCMPGTB
• VPCMPGTD
• VPCMPGTQ
• VPCMPGTW
• VPERM2I128
• VPERMD
• VPERMPD
• VPERMPS
• VPERMQ
• VPGATHERDD
• VPGATHERDQ
• VPGATHERQD
• VPGATHERQQ
• VPHADDD

All 138 AVX2 instructions

More AVX10.1 instructions

• VPADDB
• VPADDD
• VPADDQ
• VPADDSB
• VPADDSW
• VPADDUSB
• VPADDUSW
• VPADDW
• VPALIGNR
• VPANDD
• VPANDND
• VPANDNQ
• VPANDQ
• VPAVGB
• VPAVGW
• VPBLENDMB
• VPBLENDMD
• VPBLENDMQ
• VPBLENDMW
• VPBROADCASTB
• VPBROADCASTMB2Q
• VPBROADCASTMW2D
• VPBROADCASTQ
• VPBROADCASTW
• VPCLMULQDQ
• VPCMPB
• VPCMPD
• VPCMPEQB
• VPCMPEQD
• VPCMPEQQ
• VPCMPEQW
• VPCMPGTB
• VPCMPGTD
• VPCMPGTQ
• VPCMPGTW
• VPCMPQ
• VPCMPUB
• VPCMPUD
• VPCMPUQ
• VPCMPUW

All 686 AVX10.1 instructions