VSUBBF16

VSUBBF16 subtracts the bfloat16 values of its second source from those of its first, element by element, 8, 16 or 32 per register. It is the AVX10.2 counterpart of VSUBPH, with the fixed rules that VADDBF16’s page sets out: rounding to nearest even, denormal inputs and results taken as zeros, no exceptions and no MXCSR. Its opcode is 5Ch in EVEX map 5 with the 66 prefix and W0, where VSUBPH has no prefix; it takes a write mask and a broadcast from 16 bits of memory, and has no scalar form.

Differences that vanish#

Flushing results to zero breaks a rule that IEEE arithmetic keeps: with gradual underflow, x - y is zero only when x equals y. Here 00C0h minus 0080h, 1.5 × 2^-126 minus 2^-126, is 2^-127, a denormal, and VSUBBF16 returns zero. Code that guards a division with x != y can still divide by zero, which the instruction does not report either.

In our builds#

GCC 14 has no AVX10.2 option. Clang 23 at -O2 with -mavx10.2 vectorizes Z[i] = X[i] - Y[i] over global __bf16 arrays into VSUBBF16 on zmm registers with the second source from memory, but X[i] - 3 becomes VADDBF16 with -3 broadcast from a constant. A single __bf16 difference is widened to singles, subtracted with VSUBSS and narrowed with VCVTNEPS2BF16, which also reads a denormal single as zero, so that path gives zero for the example above too. No AVX10.2 processor appears in this site’s recorded cases.

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(VL // 16):
    if k1[i]:
        a = zmm2.bf16[i]
        b = src3.bf16[i]
        zmm1.bf16[i] = sub(a, b)
    elif zeroing:
        zmm1.bf16[i] = 0
    # else: zmm1.bf16[i] keeps its value
# m16bcst: src3.bf16[i] is the m16 for every i
# BF16: to nearest even, denormals flushed, no flags; MXCSR unused
# 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-E4

Attributes

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

All forms

write masking; always rounds to nearest even; denormal inputs read as zero; denormal results flushed to zero

Memory forms

embedded broadcast (tuple Full); memory fault suppression

Intrinsics

The C intrinsics of Clang 23.1.2 that compile to VSUBBF16, 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 AVX10.2 instructions

• VFNMSUB132BF16
• VFNMSUB213BF16
• VFNMSUB231BF16
• VFPCLASSBF16
• VGETEXPBF16
• VGETMANTBF16
• VMAXBF16
• VMINBF16
• VMINMAXBF16
• VMINMAXPD
• VMINMAXPH
• VMINMAXPS
• VMINMAXSD
• VMINMAXSH
• VMINMAXSS
• VMOVD
• VMOVW
• VMPSADBW
• VMULBF16
• VPDPBSSD
• VPDPBSSDS
• VPDPBSUD
• VPDPBSUDS
• VPDPBUUD
• VPDPBUUDS
• VPDPWSUD
• VPDPWSUDS
• VPDPWUSD
• VPDPWUSDS
• VPDPWUUD
• VPDPWUUDS
• VRCPBF16
• VREDUCEBF16
• VRNDSCALEBF16
• VRSQRTBF16
• VSCALEFBF16
• VSQRTBF16
• VUCOMXSD
• VUCOMXSH
• VUCOMXSS

All 96 AVX10.2 instructions