RX FPU

The FPU unit is an important part of the microcontroller because customers are increasingly confronted with a multitude of applications that require data acquisition, data modeling and subsequent fast decision process and response. The RX core and its FPU are here today to meet these demanding applications.

As shown in this diagram below, the FPU is tightly connected to the CPU sharing the same registers. As result floating point operations can be seemly performed without employing external registers reducing data transfer overhead.

Competing architecture needs to go through the extra step of loading operand values into the general registers first and then move them into the floating point’s dedicated registers. The results of the floating point unit are subsequently moved to the dedicated register and back to the general registers to be then stored into memory.

Figure 1 Competing architectures use dedicated registers for floating point operations

Additionally, the RX FPU leverage the RX’s powerful addressing mode bringing FPU’s performance even one step furthe. FPU operations can be executed taking operands both from registers, both from memory to memory and one from memory and one from register. Operation’s results can go either into register or directly into memory.

Example:

What applications benefit from having a Floating Point Unit? Generally speaking all the applications that apply mathematical model on data acquired from analog devices will benefit from having a FPU on board. Two examples shown in the video are: Motor control and thermocouple temperature conversion with Motor control results shown on the graphs below.

Figure 2 CPU Bandwidth on a Motor control application


Figure 3 Code Size in kBytes on a Motor control application