Search code, repositories, users, issues, pull requests...
Assembler is built in an optimized fashion where anything that can be precomputed, is precomputed. In the above screenshot, it's shown that an optimized build can assemble most instructions in about 15 nanoseconds, which goes down to 30 for unoptimized builds.
x64 is an array of x64Ins structs. The first member of the struct is op, or the operation, an enum defined by the asm_x64.h header. The other 4 members are x64Operand structs, which are just a combination of the type of operand with the value.
Notice the use of rax and imm(0). All x86 registers like rax (including mms, ymms etc) are defined as macros with the type x64Operand. Other types of macros:
m8(), m16(), m32(), m64(), m128(), m256() and m512() are more specific versions of the mem() macro, which references any and every size of memory for ease of use. Generally, if you know the size of memory accessed, use the size specific version of the macro that matches with the bit width of the other operands. All of these macros have the same syntax.
This is a variable length macro, with each argument being optional. Each of the register arguments of the mem() macro have to be preceeded with a $ prefix. Any 32 bit signed integer can be passed for the offset parameter, and only 1, 2, 4 and 8 are allowed in the 4th parameter, also called the "scale" parameter (ANY OTHER VALUE WILL GO TO 1, x86 limitation). The last parameter is a segment register, also preceeded with a $. Make sure to pass in $none for register parameters you are not using, as it will assume eax otherwise
