x86_64 NASM Assembly Quick Reference ("Cheat Sheet")

Instructions (basically identical to 32-bit x86)

For gory instruction set details, read the full Intel PDFs: part 1 (A-M) and part 2 (N-Z).
Mnemonic
Purpose
Examples
mov dest,src
Move data between registers, load immediate data into registers, move data between registers and memory.
mov rax,4  ; Load constant into rax
mov rdx,rax  ; Copy rax into rdx
mov rdx,[123]  ; Copy rdx to memory address 123
push src
Insert a value onto the stack.  Useful for passing arguments, saving registers, etc.
push rbp
pop dest
Remove topmost value from the stack.  Equivalent to "mov dest, [rsp]; add 8,rsp"
pop rbp
call func
Push the address of the next instruction and start executing func.
call print_int
ret
Pop the return program counter, and jump there.  Ends a subroutine.
ret
add dest,src
dest=dest+src
add rax,rdx ; Add rbx to rax
mul src
Multiply rax and src as unsigned integers, and put the result in rax.  High 64 bits of product (usually zero) go into rdx.
mul rdx ; Multiply rax by rdx
; rax=low bits, rdx overflow
div src
Divide rax by src, and put the ratio into rax, and the remainder into rdx.
Bizarrely, on input rdx must be zero, or you get a SIGFPE.
mov rdx,0 ; avoid error
div rcx ; compute rax/rcx
shr val,bits
Bitshift a value right by a constant, or the low 8 bits of rcx ("cl").
Shift count MUST go in rcx, no other register will do!
add rcx,4
shr rax,cl ; shift by rcx
jmp label Goto the instruction label:.  Skips anything else in the way. jmp post_mem
mov [0],rax ; Write to NULL!
post_mem: ; OK here...
cmp a,b

Compare two values.  Sets flags that are used by the conditional jumps (below).
cmp rax,10  
jl label Goto label if previous comparison came out as less-than.  Other conditionals available are: jle (<=), jeq (==), jge (>=), jg (>), jne (!=), and many others.
Also available in unsigned comparisons: jb (<), jbe (<=), ja (>), jae (>=).
jl loop_start  ; Jump if rax<10

Constants, Registers, Memory

"12" means decimal 12; "0xF0" is hex.  "some_function" is the address of the first instruction of the function.  Memory access (use register as pointer): "[rax]".  Same as C "*rax".
Memory access with offset (use register + offset as pointer): "[rax+4]".  Same as C "*(rax+4)".
Memory access with scaled index (register + another register * scale): "[rax+rbx*4]".  Same as C "*(rax+rbx*4)".

Registers

rax, rcx, rdx, rbx, rsp, rbp, rsi, rdi, r8, r9, r10, r11, r12, r13, r14, and r15. Registers can be accessed via a variety of names and sizes:
64-bit: rax, r8
32-bit: eax, r8d  (plus zero extension)
16-bit: ax, r8w
8-bit: al, r8b
rsp is the stack pointer
The stack frame pointer isn't used very often.
Return value in rax (or eax)
Arguments are in rdi, rsi, rdx, rcx, r8d, r9d, and then on the stack (in the usual order)
Saved registers are rsp,rbp, rbx, r12, r13, r14, and r15. All other registers can be trashed.

See sandpile.org for an opcode map.

It can be tricky figuring out if your CPU can run in 64-bit mode. Intel calls this EM64T. Pentium III's and earlier don't have EM64T. Neither do Pentium M's or even the original Core chips. Some late-model Pentium/Celeron 4/D models do have EM64T, but not very many of them (see table). All the latest Intel Core2 chips support 64-bit mode. AMD introduced "AMD64" mode in 2003; the Athlon 64, Athlon 64 FX, Opteron, and newer processors have it (most have "64" in their names); while the original Athlon, AthlonXP, and Semperons don't. In Linux, 'grep " lm " /proc/cpuinfo' will show all your CPU flags on a 64-bit ("long mode") machine; it won't show anything on a 32-bit machine.
O. Lawlor, ffosl@uaf.edu
Up to: Class Site, CS, UAF