VBROADCASTF32X4

VBROADCASTF32X4 loads four singles, 16 bytes, from memory and repeats them in every 128-bit block of a YMM or ZMM register, under a mask of singles. It is VBROADCASTI32X4 under a floating-point name, and that page covers the forms; the bits are copied, not converted, so no SIMD exception can arise. It needs AVX512F, plus AVX512VL for the YMM form, and takes only a memory source. Unmasked on YMM registers it does what AVX’s VBROADCASTF128 does, and VBROADCASTF64X2 is the same load with a mask of quadwords.

Recorded on Zen 5#

• Masked by singles. vbroadcastf32x4 zmm9{k3}, xmmword ptr [mem], with AAEBh in the low 16 bits of K3, wrote the doublewords that K3 selects, each from its place within the 16 bytes, and kept the others: element 2 kept its old value while element 3 next to it took the fourth single of memory, a split that a mask of quadwords cannot make.
• YMM. {evex} vbroadcastf32x4 ymm9, xmmword ptr [mem] wrote two copies and cleared ZMM9 from bit 256 up.

In compiled code#

GCC 14 and Clang 23 at -O2 with -march=x86-64-v4 compile _mm512_broadcast_f32x4 of a value in memory to VBROADCASTF32X4, and Clang uses it as well for _mm512_broadcast_f64x2 without a mask, where GCC keeps VBROADCASTF64X2. A loop that multiplies by a repeating pattern of four floats, g[i] = f[i] * pf[i & 3], is not vectorized by either compiler at -O2: both multiply one element at a time.

AMD’s Zen 5 latency spreadsheet gives one macro-op on pipe FP1 or FP2 for the YMM and ZMM forms.

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]:
        zmm1.d[i] = mem.d[i % 4]
    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.

All forms

AVX512-E6

Attributes

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

All forms

write masking; memory fault suppression

Intrinsics

The C intrinsics of Clang 23.1.2 that compile to VBROADCASTF32X4, 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 AVX10.1 instructions

• KXORW
• VADDPD
• VADDPH
• VADDPS
• VADDSD
• VADDSH
• VADDSS
• VAESDEC
• VAESDECLAST
• VAESENC
• VAESENCLAST
• VALIGND
• VALIGNQ
• VANDNPD
• VANDNPS
• VANDPD
• VANDPS
• VBLENDMPD
• VBLENDMPS
• VBROADCASTF32X2
• VBROADCASTF32X8
• VBROADCASTF64X2
• VBROADCASTF64X4
• VBROADCASTI32X2
• VBROADCASTI32X4
• VBROADCASTI32X8
• VBROADCASTI64X2
• VBROADCASTI64X4
• VBROADCASTSD
• VBROADCASTSS
• VCMPPD
• VCMPPH
• VCMPPS
• VCMPSD
• VCMPSH
• VCMPSS
• VCOMISD
• VCOMISH
• VCOMISS
• VCOMPRESSPD

All 686 AVX10.1 instructions