VPORD

VPORD is the EVEX form of VPOR with a write mask that selects doublewords: the bitwise OR of two 128-, 256- or 512-bit values, whose memory operand may be one doubleword repeated, {1to16} at 512 bits. VPORQ is the same with quadwords; unmasked the two give the same bits, and VPANDD shows what the EVEX forms add.

Recorded on Zen 5#

• vpord zmm14{k7}, zmm8, dword bcst [mem] with 1 in memory set bit 0 of the 10 doublewords of ZMM8 that F2E3h selects, so 80000000h became 80000001h and 10h became 11h, and left the other six of ZMM14 alone.
• vpord ymm24{k6}{z}, ymm3, ymm3 copied the three doublewords that A4h selects, an OR of a value with itself, and zeroed the rest.

In compiled code#

GCC 14 at -O2 with -march=x86-64-v4 uses VPORD for int data: d[i] = b[i] | c[i] and d[i] |= 0x80000000 over int arrays, the constant broadcast from a general register by VPBROADCASTD before the loop, and _mm512_or_epi32, _mm512_or_si512 and their masked forms. A broadcast is folded in: _mm512_or_epi32(a, _mm512_set1_epi32(*p)) is vpord zmm0, zmm0, dword ptr [rdi]{1to16}. Clang 23 with -march=x86-64-v4 uses VORPS for all of these, masked ones included; with -mavx512f it uses VPORD for the loops and the masked intrinsics and VPORQ for the unmasked ones.

Neither compiler puts the mask of a condition on the OR. For if (b[i] > 0) d[i] |= c[i];, GCC ORs all sixteen elements with VPORD and stores the result with a masked VMOVDQA32 under the mask of a VPCMPD, skipping the store when no element is selected; Clang does the same on YMM registers with the VEX VPOR.

AMD’s Zen 5 latency spreadsheet gives the EVEX VPORD one cycle and four per cycle on any of the four vector pipes. Zen 5’s 6 recorded cases cover the register and memory forms, broadcast included, at all three widths.

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] = zmm2.d[i] | src3.d[i]
    elif zeroing:
        zmm1.d[i] = 0
    # else: zmm1.d[i] keeps its value
# m32bcst: src3.d[i] is the m32 for every i
# 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

Memory forms

embedded broadcast (tuple Full); memory fault suppression

Intrinsics

The C intrinsics of Clang 23.1.2 that compile to VPORD, 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

• VPMOVWB
• VPMOVZXBD
• VPMOVZXBQ
• VPMOVZXBW
• VPMOVZXDQ
• VPMOVZXWD
• VPMOVZXWQ
• VPMULDQ
• VPMULHRSW
• VPMULHUW
• VPMULHW
• VPMULLD
• VPMULLQ
• VPMULLW
• VPMULTISHIFTQB
• VPMULUDQ
• VPOPCNTB
• VPOPCNTD
• VPOPCNTQ
• VPOPCNTW
• VPORQ
• VPROLD
• VPROLQ
• VPROLVD
• VPROLVQ
• VPRORD
• VPRORQ
• VPRORVD
• VPRORVQ
• VPSADBW
• VPSCATTERDD
• VPSCATTERDQ
• VPSCATTERQD
• VPSCATTERQQ
• VPSHLDD
• VPSHLDQ
• VPSHLDVD
• VPSHLDVQ
• VPSHLDVW
• VPSHLDW

All 686 AVX10.1 instructions