Hello world and printing numbers a few times

hello64.s

; also see an alternative: http://lpaste.net/111802

; rdi = the number (integer) to print
; prints a number as hex (basically the easy way out)
printnum:
    push    r13
    push    r14
    push    r15

    ; save the argument because we're going to use it to pass an argument to
    ; printchar (r13 is callee-save)
    mov     r13, rdi

    ; find the highest nibble
    bsr     r15, rdi    ; get most significant bit set
    and     r15, ~0x03  ; truncate to nibble granularity (ex: 0110 0111 -> 0110 0100)

.loop:
    ; fetch a byte, mask it and shift
    movzx   ecx, r15b   ; store the result in cl (from rcx)
    mov     rdi, r13
    shr     rdi, cl
    and     rdi, 0xF

    cmp     dil, 0xA
    jge     .letter

.number:
    add     dil, '0'
    jmp     .print

.letter:
    add     dil, 'A' - 0xA

.print:
    ; print the character and loop
    call    printchar

    ; decrease the variable shift and loop back if underflow
    sub     r15b, 4
    jns     .loop

    pop     r15
    pop     r14
    pop     r13

    ret