// https://github.com/nadavrot/memset_benchmark/blob/main/src/memset/impl.c

core::arch::global_asm!{r#"
.text
.globl memset
memset:
        mov     rax, rcx
        cmp     r8, 31
        ja      .LBB0_10
        cmp     r8, 4
        ja      .LBB0_5
        test    r8, r8
        je      .LBB0_26
        mov     byte ptr [rax], dl
        mov     byte ptr [r8 + rax - 1], dl
        cmp     r8, 3
        jb      .LBB0_26
        mov     byte ptr [rax + 1], dl
        mov     byte ptr [rax + 2], dl
        ret
.LBB0_10:
        movzx   ecx, dl
        movd    xmm0, ecx
        punpcklbw       xmm0, xmm0
        pshuflw xmm0, xmm0, 0
        pshufd  xmm0, xmm0, 0
        cmp     r8, 161
        jb      .LBB0_23
        movdqu  xmmword ptr [rax], xmm0
        movdqu  xmmword ptr [rax + 16], xmm0
        mov     rdx, rax
        and     rdx, -32
        lea     r9, [rax + r8]
        lea     rcx, [rax + r8]
        add     rcx, -32
        lea     r8, [rdx + 192]
        cmp     r8, r9
        jae     .LBB0_12
.LBB0_13:
        movdqa  xmmword ptr [r8 - 160], xmm0
        movdqa  xmmword ptr [r8 - 144], xmm0
        movdqa  xmmword ptr [r8 - 128], xmm0
        movdqa  xmmword ptr [r8 - 112], xmm0
        movdqa  xmmword ptr [r8 - 96], xmm0
        movdqa  xmmword ptr [r8 - 80], xmm0
        movdqa  xmmword ptr [r8 - 64], xmm0
        movdqa  xmmword ptr [r8 - 48], xmm0
        movdqa  xmmword ptr [r8 - 32], xmm0
        movdqa  xmmword ptr [r8 - 16], xmm0
        add     r8, 160
        cmp     r8, r9
        jb      .LBB0_13
        add     r8, -160
        mov     rdx, r8
        cmp     rdx, rcx
        jb      .LBB0_16
        jmp     .LBB0_17
.LBB0_5:
        cmp     r8, 16
        ja      .LBB0_9
        movzx   edx, dl
        movabs  rcx, 72340172838076673
        imul    rcx, rdx
        cmp     r8, 8
        jb      .LBB0_8
        mov     qword ptr [rax], rcx
        mov     qword ptr [rax + r8 - 8], rcx
        ret
.LBB0_23:
        lea     rcx, [rax + r8]
        add     rcx, -32
        mov     rdx, rax
.LBB0_24:
        movdqu  xmmword ptr [rdx], xmm0
        movdqu  xmmword ptr [rdx + 16], xmm0
        add     rdx, 32
        cmp     rdx, rcx
        jb      .LBB0_24
        jmp     .LBB0_25
.LBB0_12:
        add     rdx, 32
        cmp     rdx, rcx
        jae     .LBB0_17
.LBB0_16:
        movdqa  xmmword ptr [rdx], xmm0
        movdqa  xmmword ptr [rdx + 16], xmm0
        add     rdx, 32
.LBB0_17:
        cmp     rdx, rcx
        jb      .LBB0_18
        cmp     rdx, rcx
        jb      .LBB0_20
.LBB0_21:
        cmp     rdx, rcx
        jae     .LBB0_25
.LBB0_22:
        movdqa  xmmword ptr [rdx], xmm0
        movdqa  xmmword ptr [rdx + 16], xmm0
.LBB0_25:
        movdqu  xmmword ptr [rcx], xmm0
        movdqu  xmmword ptr [rcx + 16], xmm0
.LBB0_26:
        ret
.LBB0_9:
        movzx   ecx, dl
        movd    xmm0, ecx
        punpcklbw       xmm0, xmm0
        pshuflw xmm0, xmm0, 0
        pshufd  xmm0, xmm0, 0
        movdqu  xmmword ptr [rax + r8 - 16], xmm0
        movdqu  xmmword ptr [rax], xmm0
        ret
.LBB0_18:
        movdqa  xmmword ptr [rdx], xmm0
        movdqa  xmmword ptr [rdx + 16], xmm0
        add     rdx, 32
        cmp     rdx, rcx
        jae     .LBB0_21
.LBB0_20:
        movdqa  xmmword ptr [rdx], xmm0
        movdqa  xmmword ptr [rdx + 16], xmm0
        add     rdx, 32
        cmp     rdx, rcx
        jb      .LBB0_22
        jmp     .LBB0_25
.LBB0_8:
        mov     dword ptr [rax], ecx
        mov     dword ptr [rax + r8 - 4], ecx
        ret
"#}