C++ Munich Meetup - D without a runtime

main.d

int foo(int a, int b)
{
  return (a % 3235272) % b;
}

int bar(int a)
{
  return foo(a, 88);
}

extern(C) void main(int argc, char** argv) {
    auto ret = foo(argc, 88);
    version(print)
    {
        print_num(ret);
        print(argv[0]);
        print("\n");
        print(argv[1]);
    }
}

void print_num(int val)
{
    enum base = 10;
    char[11] buf = '0';
    buf[$-1] = '\n';
    auto gptr = buf.ptr + buf.length - 1;
    do {
      *(--gptr) = (val % base) + '0';
      val /= base;
    } while(val);
    print(buf.ptr, buf.length);
}

void print(const char* c)
{
    print(c, c.strlen);
}

void print(const char* c, size_t len)
{
    asm
    {
        mov RDX, len;
        mov RSI, c;
        mov RAX, 1; // write
        mov RDI, RAX;
        syscall;
    }
}

size_t strlen(const char* c)
{
    auto cptr = cast(char*) c;
    size_t i;
    while (*cptr != 0)
    {
        cptr++;
        i++;
    }
    return i;
}

output

> dmd -c -lib main.d runtime.d -of=main.o && ld.bfd main.o -o main
> perf stat -d ./main

 Performance counter stats for './main':

          0,127654      task-clock:u (msec)       #    0,191 CPUs utilized          
                 0      context-switches:u        #    0,000 K/sec                  
                 0      cpu-migrations:u          #    0,000 K/sec                  
                 1      page-faults:u             #    0,008 M/sec                  
             1.489      cycles:u                  #    0,012 GHz                    
                39      instructions:u            #    0,03  insn per cycle         
                 7      branches:u                #    0,055 M/sec                  
                 0      branch-misses:u           #    0,00% of all branches        
                10      L1-dcache-loads:u         #    0,078 M/sec                  
                11      L1-dcache-load-misses:u   #  110,00% of all L1-dcache hits  
     <not counted>      LLC-loads:u                                                   (0,00%)
     <not counted>      LLC-load-misses:u                                             (0,00%)

       0,000667675 seconds time elapsed

> ldc -O3 -c -lib main.d runtime.d -of=main.a && ld.bfd main.a -o main
> perf stat -d ./main

 Performance counter stats for './main':

          0,130168      task-clock:u (msec)       #    0,154 CPUs utilized          
                 0      context-switches:u        #    0,000 K/sec                  
                 0      cpu-migrations:u          #    0,000 K/sec                  
                 1      page-faults:u             #    0,008 M/sec                  
             1.467      cycles:u                  #    0,011 GHz                    
                18      instructions:u            #    0,01  insn per cycle         
                 5      branches:u                #    0,038 M/sec                  
                 0      branch-misses:u           #    0,00% of all branches        
                 6      L1-dcache-loads:u         #    0,046 M/sec                  
                10      L1-dcache-load-misses:u   #  166,67% of all L1-dcache hits  
     <not counted>      LLC-loads:u                                                   (0,00%)
     <not counted>      LLC-load-misses:u                                             (0,00%)

       0,000846418 seconds time elapsed
       
> ldc -d-version=print -betterC -O3 -c -lib main.d runtime.d -of=main.a && ld.bfd main.a -o main && ./main 123 456
0000000003
./main
123

runtime.d

extern(C) void __d_sys_exit(long arg1)
{
    asm
    {
        mov RAX, 60;
        mov RDI, arg1;
        syscall;
    }
}

extern extern(C) void main(int, char**);
private extern(C) void _start()
{
    int argc;
    char** argv;
    asm
    {
        mov RCX, 24[RSP];
        mov argc, ECX;
        lea RCX, 32[RSP];
        mov argv, RCX;
    }
    main(argc, argv);
    __d_sys_exit(0);
}