[REQUEST] Missing vcpmleps instruction on SKX

[ivan-pi]

Is your feature request related to a problem? Please describe.

I can't analyze the following loop (generated by the Intel ifx compiler):

.LBB0_112:
        vmovups -8(%r10,%r14,4), %ymm6
        vmovups -4(%r10,%r14,4), %ymm7
        vaddps  (%r10,%r14,4), %ymm6, %ymm6
        vaddps  (%r11,%r14,4), %ymm6, %ymm6
        vaddps  4(%r12,%r14,4), %ymm6, %ymm6
        vmulps  %ymm2, %ymm6, %ymm6
        vsubps  %ymm6, %ymm7, %ymm7
        vandps  %ymm3, %ymm7, %ymm7
        vcmpleps        %ymm5, %ymm7, %ymm8
        vblendvps       %ymm8, %ymm5, %ymm7, %ymm5
        vmovups %ymm6, (%r9,%r14,4)
        addq    $8, %r14
        cmpq    %r8, %r14
        jl      .LBB0_112

The performance data for the vcpmleps instruction is missing.

Describe the solution you'd like

The data were available, at least for SKX and SPR.

Additional context

Link to Compiler Explorer for the cause under scrutiny: https://godbolt.org/z/5h5zxr6o7