vec8

julia> code_native(add,(Vec{8,Float32},Vec{8,Float32}))
    .text
Filename: none
Source line: 0
    pushq   %rbp
    movq    %rsp, %rbp
Source line: 2
    vaddss  %xmm1, %xmm0, %xmm8
    vmovshdup   %xmm0, %xmm3    # xmm3 = xmm0[1,1,3,3]
    vmovshdup   %xmm1, %xmm4    # xmm4 = xmm1[1,1,3,3]
    vaddss  %xmm4, %xmm3, %xmm9
    vpermilpd   $1, %xmm0, %xmm4 # xmm4 = xmm0[1,0]
    vpermilpd   $1, %xmm1, %xmm5 # xmm5 = xmm1[1,0]
    vaddss  %xmm5, %xmm4, %xmm4
    vpermilps   $231, %xmm0, %xmm5 # xmm5 = xmm0[3,1,2,3]
    vpermilps   $231, %xmm1, %xmm6 # xmm6 = xmm1[3,1,2,3]
    vaddss  %xmm6, %xmm5, %xmm5
    vextractf128    $1, %ymm0, %xmm0
    vextractf128    $1, %ymm1, %xmm1
    vaddss  %xmm1, %xmm0, %xmm6
    vmovshdup   %xmm0, %xmm7    # xmm7 = xmm0[1,1,3,3]
    vmovshdup   %xmm1, %xmm2    # xmm2 = xmm1[1,1,3,3]
    vaddss  %xmm2, %xmm7, %xmm2
    vpermilpd   $1, %xmm0, %xmm7 # xmm7 = xmm0[1,0]
    vpermilpd   $1, %xmm1, %xmm3 # xmm3 = xmm1[1,0]
    vaddss  %xmm3, %xmm7, %xmm3
    vpermilps   $231, %xmm0, %xmm0 # xmm0 = xmm0[3,1,2,3]
    vpermilps   $231, %xmm1, %xmm1 # xmm1 = xmm1[3,1,2,3]
    vaddss  %xmm1, %xmm0, %xmm0
    vinsertps   $16, %xmm2, %xmm6, %xmm1 # xmm1 = xmm6[0],xmm2[0],xmm6[2,3]
    vinsertps   $32, %xmm3, %xmm1, %xmm1 # xmm1 = xmm1[0,1],xmm3[0],xmm1[3]
    vinsertps   $48, %xmm0, %xmm1, %xmm0 # xmm0 = xmm1[0,1,2],xmm0[0]
    vinsertps   $16, %xmm9, %xmm8, %xmm1 # xmm1 = xmm8[0],xmm9[0],xmm8[2,3]
    vinsertps   $32, %xmm4, %xmm1, %xmm1 # xmm1 = xmm1[0,1],xmm4[0],xmm1[3]
    vinsertps   $48, %xmm5, %xmm1, %xmm1 # xmm1 = xmm1[0,1,2],xmm5[0]
    vinsertf128 $1, %xmm0, %ymm1, %ymm0
    vmovaps %ymm0, (%rdi)
    movq    %rdi, %rax
    popq    %rbp
    vzeroupper
    retq