The other one, is a clunky one from UNIX world point of view on x86 CPUs, follow the links on the sibling comment.
Rather clunky and most UNIX Assemblers were never as powerfull as other systems macro assemblers, since after UNIX System V, it was there only as an additional external tool pass for the C compilation steps, and even before that, the assembler was quite minimalist.
Then there is the whole brain twist of being used to op dest, src, and having to switch into op src, dest.
I prefer the Intel approach as it is more similar to dest = src, rather than src -> dest.
Intel's approach is also more common across various assembly languages.
I think the schizophrenic notation 12(%ebp) is much worse than the dest/src switch.
dest/src order is not a big deal in absolute terms: The ISA designer picks an order, documents it, everyone follows their documentation, done. Intel picks one order for programming Intel, DEC picks another order for programming the VAX? Great, just look in the instruction set manual and you're set.
The silly thing is importing one order to another architecture that uses the opposite order. Now your users have to transpose operands for no reason. The really silly thing is messing up the operand order of the cmp instruction in the process. The VAX assembly language had a compare with the sane operand order, BTW.
Yes, that is the cherry on top, maybe.