main.cpp

#include "stdio.h"
#include <math.h>

extern "C" bool Cc4(const double* ht, const double* wt, int n, double* bsa1, double* bsa2, double* bsa3);

int main(int argc, char* argv[])
{
    const int n = 6;
    const double ht[n] = { 150, 160, 170, 180, 190, 200 };
    const double wt[n] = { 50.0, 60.0, 70.0, 80.0, 90.0, 100.0 };
    double bsa1_a[n], bsa1_b[n];
    double bsa2_a[n], bsa2_b[n];
    double bsa3_a[n], bsa3_b[n];

    for (int i = 0; i < n; i++)
    {
        bsa1_a[i] = 0.007184 * pow(ht[i], 0.725) * pow(wt[i], 0.425);
        bsa2_a[i] = 0.0235 * pow(ht[i], 0.42246) * pow(wt[i], 0.51456);
        bsa3_a[i] = sqrt(ht[i] * wt[i]) / 60.0;
    }

    Cc4(ht, wt, n, bsa1_b, bsa2_b, bsa3_b);

    printf("Results for CallingConvention4\n\n");

    for (int i = 0; i < n; i++)
    {
        printf("height: %6.1lf cm\n", ht[i]);
        printf("weight: %6.1lf kg\n", wt[i]);
        printf("BSA (C++):    %10.6lf %10.6lf %10.6lf (sq. m)\n", bsa1_a[i], bsa2_a[i], bsa3_a[i]);
        printf("BSA (X86-64): %10.6lf %10.6lf %10.6lf (sq. m)\n", bsa1_b[i], bsa2_b[i], bsa3_b[i]);
        printf("\n");
    }
    return 0;
}

callingconvention.asm

; Name:     callingconvention4.asm
;
; Build:    g++ -c main.cpp -o main.o
;           nasm -f elf64 -o callingconvention4.o callingconvention4.asm
;           g++ -o callingconvention4 callingconvention4.o main.o
;
; Source:   Modern x86 Assembly Language Programming p.540

bits 64

extern pow

global Cc4

section .data
; Floating-point constants for BSA equations

r8_0p007184:    dq 0.007184
r8_0p725:       dq 0.725
r8_0p425:       dq 0.425
r8_0p0235:      dq 0.0235
r8_0p42246:     dq 0.42246
r8_0p51456:     dq 0.51456
r8_60p0:        dq 60.0

section .text

; extern "C" bool Cc4(const double* ht, const double* wt, int n, double* bsa1, double* bsa2, double* bsa3);
;

Cc4:
; Registers: rdi    ht
;            rsi    wt
;            rdx    n
;            rcx    bsa1
;            r8     bsa2
;            r9     bsa3

push    rbp
mov     rbp,rsp
sub     rsp,8                           ;align stack
push    r12                             ;save non-volatile register
push    r13
push    r14
push    r15
push    rbx

; Initialize registers unchanged by pow function
xor     eax,eax                         ;assume error
test    edx,edx                         ;is n <= 0?
jle     .done                           ;jump if n <= 0
xor     r10,r10                         ;array element offset
mov     r11,rdi                         ;height
mov     r12,rsi                         ;width
mov     r13,rcx                         ;bsa1
mov     r14,r8                          ;bsa2
mov     r15,r9                          ;bsa3
;register r9 is unchanged by pow function
mov     r9,rdx                              ;n

; Registers: r9     n
;            r10    array element offset
;            r11    ht
;            r12    wt
;            r13    bsa1
;            r14    bsa2
;            r15    bsa3

; Calculate bsa1 = 0.007184 * pow(ht, 0.725) * pow(wt, 0.425);
.l1:

;xmm0 = 0.007184
movsd   xmm0,[r8_0p007184]
movsd   [r13+r10],xmm0

;xmm0 = 0.007184 * pow(ht, 0.725)
movsd   xmm0,[r11+r10]                  ;xmm0 = height
movsd   xmm1,[r8_0p725]
call    pow
movsd   xmm1,[r13+r10]
mulsd   xmm0,xmm1
movsd   [r13+r10],xmm0

;xmm0 = 0.007184 * pow(ht, 0.725) * pow(wt, 0.425)
movsd   xmm0,qword[r12+r10]
movsd   xmm1,qword[r8_0p425]
call    pow
movsd   xmm1,[r13+r10]
mulsd   xmm0,xmm1
movsd   [r13+r10],xmm0

; Calculate bsa2 = 0.0235 * pow(ht, 0.42246) * pow(wt, 0.51456)

;xmm0 = 0.0235
movsd   xmm0,[r8_0p0235]
movsd   [r14+r10],xmm0

;xmm0 = 0.0235 * pow(ht, 0.42246)
movsd   xmm0,[r11+r10]                  ;xmm0 = height
movsd   xmm1,[r8_0p42246]
call    pow
movsd   xmm1,[r14+r10]
mulsd   xmm0,xmm1
movsd   [r14+r10],xmm0

;xmm0 = 0.0235 * pow(ht, 0.42246) * pow(wt, 0.51456)
movsd   xmm0,qword[r12+r10]
movsd   xmm1,qword[r8_0p51456]
call    pow
movsd   xmm1,[r14+r10]
mulsd   xmm0,xmm1
movsd   [r14+r10],xmm0

; Calculate bsa3 = sqrt(ht * wt) / 60.0;
movq    xmm0,[r11+r10]
movq    xmm1,[r12+r10]
mulsd   xmm0,xmm1
sqrtsd  xmm0,xmm0
divsd   xmm0,[r8_60p0]                  ;xmm0 = bsa3

movsd   [r15+r10],xmm0                  ;save bsa3 result

add     r10,8                           ;update array offset
dec     r9                              ;n = n - 1
jnz     .l1
mov     eax,1                           ;set success return code

.done:
pop     rbx
pop     r15
pop     r14
pop     r13
pop     r12
mov     rsp,rbp
pop     rbp
ret

build

g++ -c main.cpp -o main.o
nasm -f elf64 -o callingconvention4.o callingconvention4.asm
g++ -o callingconvention4 callingconvention4.o main.o