Today TI announced that it had created the industry's first floating point digital signal controllers, the F2338x family. This claim sounded wrong to me. I was certain that Infineon had offered floating-point digital signal controllers (also called DSCs) for several years. And sure enough, a quick search turned up the floating-point TC1130 family.
However, a little more digging revealed that TI's claim is technically correct. Infineon calls the TC1130 family "microcontrollers," not "DSCs." By this logic, TI does in fact have the first floating-point DSCs. On the other hand, the TC1130 and F2338x families have many similarities, and they clearly compete with each other. Thus, it is silly to pretend that TI has developed a ground-breaking technology.
The basic problem here is that the term "DSC" is poorly defined. Depending on the vendor, the terms DSC, DSP, and microcontroller can all mean the same thing. In light of this problem, I propose defining DSCs as chips that offers the following features:
- Single-cycle multipliers
- Flash memory
- ADCs and PWMs
It's worth noting that this naming ambiguity is not limited to DSCs. For example, Analog Devices doesn't call its DSP/MCU Blackfin platform either a DSP or a MCU. Instead, ADI uses the completely made-up term "convergent processor."
The point to all of this is that you shouldn't get distracted by vendor's sometimes hair-brained naming schemes. Instead, you should just focus on what you need the processor to do.