As you may have noticed (I hope you have), the procedures discussed in the previous section do not receive any parameters at all. Therefore, we do need to supply two more procedures in order to make it possible to tell the core where data resides and how many bytes to process. As each of these procedures accepts one parameter, the code would be a bit more segmented in order to reflect the calling convention being used, which, in our case, is as follows:

• cdecl for 32-bit targets
• Microsoft x64 for Windows-based 64-bit targets
• AMD64 for Linux-based 64-bit targets