x86-asm-dot-full.s

.section .text,"xr",one_only,blogpost::avx2_dot::avx2_dot_product
        .globl  blogpost::avx2_dot::avx2_dot_product
        .p2align        4, 0x90
blogpost::avx2_dot::avx2_dot_product:
        .cv_func_id 6
.seh_proc _ZN8blogpost8avx2_dot16avx2_dot_product17hffb9005f074b96fbE
        sub rsp, 104
        .seh_stackalloc 104
        .seh_endprologue
        mov qword ptr [rsp + 40], rdx
        mov qword ptr [rsp + 48], r9
        cmp rdx, r9
        jne .LBB4_16
        mov rax, rdx
        and rax, -8
        je .LBB4_2
        lea r10, [rax - 1]
        shr r10, 3
        inc r10
        mov r9d, r10d
        and r9d, 7
        cmp rax, 57
        jae .LBB4_14
        vxorps xmm0, xmm0, xmm0
        xor eax, eax
        jmp .LBB4_5
.LBB4_2:
        xor eax, eax
        vxorps xmm0, xmm0, xmm0
        jmp .LBB4_7
.LBB4_14:
        and r10, -8
        vxorps xmm0, xmm0, xmm0
        xor eax, eax
        .p2align        4, 0x90
.LBB4_15:
        .cv_inline_site_id 7 within 6 inlined_at 6 24 0
        .cv_inline_site_id 8 within 7 inlined_at 8 1465 0
        vmovups ymm1, ymmword ptr [rcx + 4*rax]
        vmovups ymm2, ymmword ptr [rcx + 4*rax + 32]
        vmovups ymm3, ymmword ptr [rcx + 4*rax + 64]
        vmovups ymm4, ymmword ptr [rcx + 4*rax + 96]
        .cv_inline_site_id 9 within 6 inlined_at 6 27 0
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax]
        .cv_inline_site_id 10 within 6 inlined_at 6 28 0
        vaddps ymm0, ymm0, ymm1
        vmulps ymm1, ymm2, ymmword ptr [r8 + 4*rax + 32]
        vaddps ymm0, ymm0, ymm1
        vmulps ymm1, ymm3, ymmword ptr [r8 + 4*rax + 64]
        vaddps ymm0, ymm0, ymm1
        vmulps ymm1, ymm4, ymmword ptr [r8 + 4*rax + 96]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 128]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 128]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 160]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 160]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 192]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 192]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 224]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 224]
        add rax, 64
        add r10, -8
        vaddps ymm0, ymm0, ymm1
        jne .LBB4_15
.LBB4_5:
        test r9, r9
        je .LBB4_7
        .p2align        4, 0x90
.LBB4_6:
        vmovups ymm1, ymmword ptr [rcx + 4*rax]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax]
        add rax, 8
        dec r9
        vaddps ymm0, ymm0, ymm1
        jne .LBB4_6
.LBB4_7:
        .cv_inline_site_id 11 within 6 inlined_at 6 42 0
        .cv_inline_site_id 12 within 11 inlined_at 6 61 0
        vextractf128 xmm1, ymm0, 1
        mov r9, rax
        sub r9, rdx
        .cv_inline_site_id 13 within 11 inlined_at 6 63 0
        vaddps xmm0, xmm1, xmm0
        .cv_inline_site_id 14 within 11 inlined_at 6 66 0
        vshufpd xmm1, xmm0, xmm0, 1
        .cv_inline_site_id 15 within 11 inlined_at 6 67 0
        vaddps xmm0, xmm0, xmm1
        .cv_inline_site_id 16 within 11 inlined_at 6 71 0
        vmovshdup xmm1, xmm0
        vaddss xmm0, xmm0, xmm1
        jae .LBB4_13
        mov r10d, edx
        sub r10d, eax
        and r10d, 7
        je .LBB4_10
        .p2align        4, 0x90
.LBB4_9:
        vmovss xmm1, dword ptr [rcx + 4*rax]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax]
        inc rax
        dec r10
        vaddss xmm0, xmm0, xmm1
        jne .LBB4_9
.LBB4_10:
        cmp r9, -8
        ja .LBB4_13
        neg rdx
        add rax, 7
        .p2align        4, 0x90
.LBB4_12:
        vmovss xmm1, dword ptr [rcx + 4*rax - 28]
        vmovss xmm2, dword ptr [rcx + 4*rax - 24]
        lea r9, [rdx + rax + 8]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 28]
        vaddss xmm0, xmm0, xmm1
        vmulss xmm1, xmm2, dword ptr [r8 + 4*rax - 24]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 20]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 20]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 16]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 16]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 12]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 12]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 8]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 8]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 4]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 4]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax]
        add rax, 8
        vaddss xmm0, xmm0, xmm1
        cmp r9, 7
        jne .LBB4_12
.LBB4_13:
        add rsp, 104
        vzeroupper
        ret
.LBB4_16:
        lea rax, [rip + __unnamed_2]
        lea rcx, [rip + __unnamed_3]
        lea r9, [rip + __unnamed_5]
        lea rdx, [rsp + 48]
        lea r8, [rsp + 56]
        vxorps xmm0, xmm0, xmm0
        mov qword ptr [rsp + 56], rax
        mov qword ptr [rsp + 64], 1
        mov qword ptr [rsp + 72], rcx
        lea rcx, [rsp + 40]
        vmovups xmmword ptr [rsp + 80], xmm0
        call core::panicking::assert_failed
        int3

x86-asm-dot-simple.s

blogpost::avx2_dot::avx2_dot_product:
        sub rsp, 104
        mov qword ptr [rsp + 40], rdx
        mov qword ptr [rsp + 48], r9
        cmp rdx, r9
        jne .LBB4_16
        mov rax, rdx
        and rax, -8
        je .LBB4_2
        lea r10, [rax - 1]
        shr r10, 3
        inc r10
        mov r9d, r10d
        and r9d, 7
        cmp rax, 57
        jae .LBB4_14
        vxorps xmm0, xmm0, xmm0
        xor eax, eax
        jmp .LBB4_5
.LBB4_2:
        xor eax, eax
        vxorps xmm0, xmm0, xmm0
        jmp .LBB4_7
.LBB4_14:
        and r10, -8
        vxorps xmm0, xmm0, xmm0
        xor eax, eax
.LBB4_15:
        vmovups ymm1, ymmword ptr [rcx + 4*rax]
        vmovups ymm2, ymmword ptr [rcx + 4*rax + 32]
        vmovups ymm3, ymmword ptr [rcx + 4*rax + 64]
        vmovups ymm4, ymmword ptr [rcx + 4*rax + 96]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax]
        vaddps ymm0, ymm0, ymm1
        vmulps ymm1, ymm2, ymmword ptr [r8 + 4*rax + 32]
        vaddps ymm0, ymm0, ymm1
        vmulps ymm1, ymm3, ymmword ptr [r8 + 4*rax + 64]
        vaddps ymm0, ymm0, ymm1
        vmulps ymm1, ymm4, ymmword ptr [r8 + 4*rax + 96]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 128]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 128]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 160]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 160]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 192]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 192]
        vaddps ymm0, ymm0, ymm1
        vmovups ymm1, ymmword ptr [rcx + 4*rax + 224]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax + 224]
        add rax, 64
        add r10, -8
        vaddps ymm0, ymm0, ymm1
        jne .LBB4_15
.LBB4_5:
        test r9, r9
        je .LBB4_7
        .p2align        4, 0x90
.LBB4_6:
        vmovups ymm1, ymmword ptr [rcx + 4*rax]
        vmulps ymm1, ymm1, ymmword ptr [r8 + 4*rax]
        add rax, 8
        dec r9
        vaddps ymm0, ymm0, ymm1
        jne .LBB4_6
.LBB4_7:
        .cv_inline_site_id 11 within 6 inlined_at 6 42 0
        .cv_inline_site_id 12 within 11 inlined_at 6 61 0
        vextractf128 xmm1, ymm0, 1
        mov r9, rax
        sub r9, rdx
        .cv_inline_site_id 13 within 11 inlined_at 6 63 0
        vaddps xmm0, xmm1, xmm0
        .cv_inline_site_id 14 within 11 inlined_at 6 66 0
        vshufpd xmm1, xmm0, xmm0, 1
        .cv_inline_site_id 15 within 11 inlined_at 6 67 0
        vaddps xmm0, xmm0, xmm1
        .cv_inline_site_id 16 within 11 inlined_at 6 71 0
        vmovshdup xmm1, xmm0
        vaddss xmm0, xmm0, xmm1
        jae .LBB4_13
        mov r10d, edx
        sub r10d, eax
        and r10d, 7
        je .LBB4_10
.LBB4_9:
        vmovss xmm1, dword ptr [rcx + 4*rax]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax]
        inc rax
        dec r10
        vaddss xmm0, xmm0, xmm1
        jne .LBB4_9
.LBB4_10:
        cmp r9, -8
        ja .LBB4_13
        neg rdx
        add rax, 7
.LBB4_12:
        vmovss xmm1, dword ptr [rcx + 4*rax - 28]
        vmovss xmm2, dword ptr [rcx + 4*rax - 24]
        lea r9, [rdx + rax + 8]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 28]
        vaddss xmm0, xmm0, xmm1
        vmulss xmm1, xmm2, dword ptr [r8 + 4*rax - 24]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 20]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 20]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 16]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 16]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 12]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 12]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 8]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 8]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax - 4]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax - 4]
        vaddss xmm0, xmm0, xmm1
        vmovss xmm1, dword ptr [rcx + 4*rax]
        vmulss xmm1, xmm1, dword ptr [r8 + 4*rax]
        add rax, 8
        vaddss xmm0, xmm0, xmm1
        cmp r9, 7
        jne .LBB4_12
.LBB4_13:
        add rsp, 104
        vzeroupper
        ret
.LBB4_16:
        lea rax, [rip + __unnamed_2]
        lea rcx, [rip + __unnamed_3]
        lea r9, [rip + __unnamed_5]
        lea rdx, [rsp + 48]
        lea r8, [rsp + 56]
        vxorps xmm0, xmm0, xmm0
        mov qword ptr [rsp + 56], rax
        mov qword ptr [rsp + 64], 1
        mov qword ptr [rsp + 72], rcx
        lea rcx, [rsp + 40]
        vmovups xmmword ptr [rsp + 80], xmm0
        call core::panicking::assert_failed
        int3