#include "asm.h"

int getSum(int ar[], int n) {
    asm("movl $0, %%eax;\
       movl %[n], %%ecx;\
       movl $0, %%edx;\
       cmp %%eax, %%ecx;\
       jle pexit1;\
       lp:;\
       dec %%ecx;\
       addl (%[ar], %%rcx, 4), %%eax;\
       cmp %%edx, %%ecx;\
       jg lp;\
       pexit1:;"
    :
    : [ar]"r"(ar), [n]"r"(n)
    : "eax", "ecx", "edx"
    );
}

int getSumFastSSE(int *ar, int n) {
    asm("movl %[n], %%ecx;\
       xorpd %%xmm0, %%xmm0;\
       movl $0, %%eax;\
       cmp $3, %%ecx;\
       jle pexit31;\
       subq $4, %%rcx;\
       lp3:\
       movdqu (%[ar], %%rcx, 4), %%xmm1;\
       paddd %%xmm1, %%xmm0;\
       subq $4, %%rcx;\
       jge lp3;\
       phaddd %%xmm0, %%xmm0;\
       phaddd %%xmm0, %%xmm0;\
       movq %%xmm0, %%rax;\
       addq $4, %%rcx;\
       pexit31:;\
       cmp $0, %%ecx;\
       jle pexit3;\
       lp4:\
       addl -4(%[ar], %%rcx, 4), %%eax;\
       dec %%rcx;\
       jnz lp4;\
       pexit3:;"
    :
    : [ar]"r"(ar), [n]"r"(n)
    :  "rcx", "xmm0", "rax", "xmm1"
    );
}

int getSumFastAVX(int *ar, int n) {
    asm("movl %[n], %%ecx;\
      vxorpd %%ymm0, %%ymm0, %%ymm0;\
      movl $0, %%eax;\
      cmp $7, %%ecx;\
      jle pexit71;\
      subq $8, %%rcx;\
      lp7:;\
      vpaddd (%[ar], %%rcx, 4), %%ymm0, %%ymm0;\
      subq $8, %%rcx;\
      jge lp7;\
      addq $8, %%rcx;\
      pexit71:;\
      cmp $0, %%ecx;\
      jle pexit7;\
      lp71:\
      addl -4(%[ar], %%rcx, 4), %%eax;\
      dec %%rcx;\
      jnz lp71;\
      pexit7:;\
      push %%rbp;\
      mov %%rsp, %%rbp;\
      subq $32, %%rsp;\
      vmovdqu %%ymm0, (%%rsp);\
      movl $8, %%ecx;\
      lp72:\
      addl -4(%%rsp, %%rcx, 4), %%eax;\
      dec %%ecx;\
      jnz lp72;\
      mov %%rbp, %%rsp;\
      pop %%rbp;\
      ret;"
    :
    : [ar]"r"(ar), [n]"r"(n)
    :  "rcx", "xmm0", "rax", "xmm1"
    );
}

double getSum(double ar[], int n) {
    asm("movl %[n], %%ecx;\
       xorpd   %%xmm0, %%xmm0;\
       movl $0, %%eax;\
       cmp %%eax, %%ecx;\
       jle pexit2;\
       lp1:\
       addsd -8(%[ar], %%rcx, 8), %%xmm0;\
       loop lp1;\
       movq %%xmm0, %%rax;\
       pexit2:;"
    :
    : [ar]"r"(ar), [n]"r"(n)
    :  "ecx", "xmm0", "rax"
    );
}