:

Non-intel compile?

Added by: ~Andrew, 2016 X 19

Added by: ~Andrew, 2016 X 19

I'm attempting to build a project that uses ttmath on IBM Power, and I'm running into an issue.

I've added the #define TTMATH_NOASM directive in an attempt to get the library to to attempt to include the x85 assembler in favor of the noasm versions of the template. But I'm still running into ambiguous overload errors and unknown conversion errors.

I apologize for the newbie questions, but how do I properly set directives to allow this code to build on a system that isn't an x86?

THanks!

Added by: tomek, 2016 X 21

> I've added the #define TTMATH_NOASM directive in a

Yes, this will use only pure C++ without any asm code.

> But I'm still running into ambiguous overload errors and unknown conversion errors.

This seems to be a compilation problem, could you show a small example (code and errors).

> I apologize for the newbie questions, but how do I properly set directives to allow this code to build on a system that isn't an x86?

I never tested this library on anything different from x86 and amd64, so you do it on your own. Before beginning make sure that the processor works in little endian mode.

Added by: ~Atul, 2016 XII 16

The issue here is not related to ttmath but the application (voltdb). A sample code doing very same operations works just fine - I verified that. So I guess the solution lies in the domain of VoltDB instead of ttmath. Hope this helps! Thanks!