mapping of the floating point registers

[xiaywang]

Hello,
I had to use both vega and mr wolf at the same time. I installed only one toolchain, follwing the instructions on vega sdk repo which lead me here.
The code compiled with this toolchain for mr wolf gave completely wrong results for logf(). The reason is that the conversions between floating point registers and integer registers, which are necessary for mr wolf, were not done with this toolchain.
Best,
Xia