Erik:
If the code is entirely in assembler source and is all absolutely located then I would suspect that the codes would be nearly the same at the target object code level. There could be some differences however for the following reasons...

1) The two assemblers could optimize the use of the CALL and JMP opcodes differently.

2) Some assemblers may actually optimize an LCALL or an LJMP to a shorter version based upon the settings of command line options or environment variables.

3) If multiple objects are linked and the target run time locations of code and/or data ojbects are determined by the linker there is surely a good possibility that the final object images could be different.

4) There are probably other issues to consider as well that relate to various assembler built-in macros that may end up generating different code on the two different platforms.

My recommendation is that you go ahead and use the Keil full platform for code generation, test and debug. Then just before final delivery that you compile with the Meta tool set and run a software validation test on the final image to ensure program functions properly....that is unless you see that the objects end up identical which you can perform with a straightforward file compare at the binary image level.

Michael Karas