Tool/software: TI C/C++ Compiler
Hi all,
I have this question for a while.
On the devices(C28x+FPU) that do not support TMU, one way to improve the float point operations is to use the FastRTS lib routines.
And it is also true that TI would add more routines under FastRTS in coming days.
Now when we come to devices like F280049C with FPU+TMU, it is true that TMU supports most of the FastRTS routines.
Definitely we can get an edge over FastRTS by using TMU.
But here are my 2 questions.
1. Does TMU support or plan to support all the FastRTS routines? If that is case then I see it useful to just enable TMU and forget about FastRTS on devices like F280049C.
2. But if TMU supports only subset of FastRTS then we would still need FastRTS for the routines not supported via TMU. In that case can we use both TMU and FastRTS at the sametime? If yes, then which one will take priority in linking? Will the code generation tool handle this smartly to replace TMU routines first and then FastRTS routines where TMU is not supported?
