I suspect that what you say is true. One thing worth considering though is that on x86_64 you actually use the vector register file for floating point (and it c

I suspect that what you say is true. One thing worth considering though is that on x86_64 you actually use the vector register file for floating point (and it can do integer too!). On the MRISC32 you have a vector register file with 32 registers that can easily be configured to do scalar operations (integer or floating point) by setting the vector length to 1.