I'm trying to understand suffixes used for the AT&T syntax for the x64 assembly used for instruction mnemonics.

For regular cases:

'b',     // 8_bit
'w',     // 16_bit
'l',     // 32_bit
'q',     // 64_bit
't',     // 80_bit

Examples: Intel vs AT&T:

inc word ptr [rbx]        incw   (%rbx)
inc dword ptr [rbx]       incl   (%rbx)
inc qword ptr [rbx]       incq   (%rbx)

But then for floating-point instructions, it's different:

L'b',     // 8_bit
L'w',     // 16_bit
L's',     // 32_bit
L'l',     // 64_bit
L't',     // 80_bit

If so, then why does the GCC compiler give me this:

fiadd word ptr [rcx]           fiadds (%rcx)
fiadd dword ptr [rcx]          fiaddl (%rcx)

Can some confirm if I'm right, or correct me?

1 Answer 1


Yes, you are right: Operation suffixes on wikibooks.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.