loop

    4.4
s* 17.5

--mul
r 13.1 
e 18.62
x 4.4 //bmi2
s* 17.5

--div
r 379 //125
e 114 //94
ss 48.35
sd 57

--vpslld
4.4

.global _k
_k:
movl    %edi, %eax
shll    $32, %eax
mov %xmm0, %rax
addl %ecx, (%eax)
ret

.global _t
_t:
    mov $2, %rdx
    mov $0xf1234567, %rcx
aa:
 mulx %rcx, %rdi, %rsi
 loop aa
 ret

 .global _f0
 _f0:
 ret

#define A() addss %xmm0, %xmm0
 //mulx %rcx, %rsi, %rdi
.global _kg
  _kg:
     mov $0xf1234567, %rcx
     mov $0, %rdx
   a0:
  //vpaddd  %ymm0, %ymm0, %ymm0
//vpslld  $20, %xmm0, %xmm0
  //div %rcx
  //shr $1, %rax
  //divsd %xmm0, %xmm0
  //add %rax, %rax
    //add %r8, %r8
    //add %r9, %r9
    //add %rsi, %rsi
    //add %rdi, %rsi //4 loop is better

    A()
    A()
    A()
    A()
    //A()
    sub $1, %rcx
    jnz a0 //loop too slow
  ret