Eye Current

A Matter of Scale.

I stumbled across an article that was promoting 16 bit architectures, In automobile control units. In my naivete I assumed it was a discussion of 8 bit over 16 bit. Struck me as a ironic because this discussion was going on 20 years ago. I was presumptuous:

In many instances, adhering to these criteria leaves automotive system developers with a handful of options, usually a mixed bag of 16-bit and 32-bit MCUs. That’s when the question of the number of bits is worth considering, and not just in the first stage of the screening process. MCU bit size indeed matters when looking at the bigger picture of research and development spending versus production volumes and system scalability requirements (increasing or decreasing) in future evolutions.

Today’s 32-bit MCU architectures provide a number of benefits to developers, beginning with an inherently high performance core, a broad linear address space and native 32-bit arithmetic capabilities. These architectures are very attractive to software designers who enjoy the convenience and headroom of a 32-bit address space at their disposal. All these benefits certainly help with software reuse and software platform design. But as every designer knows, there is no such thing as a “free lunch” in embedded systems development. So what are the drawbacks to software-friendly architectures?

What Manuel Alves of Freescale Semiconductor is suggesting is that sometimes 16 bit proceessors might be more apropriate than 32 bit processors.

I was working with some folks in Freescale and they were telling that their biggest focus and market is in the automobile industry. Todays automobiles have processors that not only control control egine performance, but braking and stability systems are commonplace. New devices such as the I-Phone will typically only have one processor while the automobile can have 6, 7 0r more processing systems.

posted by Ray L. @ 6:08 AM