erik malund wrote:
-------------------------------
... bank support ... overhead

Why screw around with banking when there is the Philips MX and offerings from other vendors that address multimegabytes without banking?


Erik,

In general I'd agree with you. The effort it took to get our project working under the banked memory model as well as the overhead (execution cycles, RAM/stack space, and incremental code space) would definitely be very nice to avoid if a better alternative were evident. Unfortunately in light of the specialty feature of the microcontroller we are using (on-screen video) and the point in the development cycle we are in (hardware locked-in, firmware nearing completion) the most prudent path seems to be to march forward with banking. This experience has put the issue on the table and we're actively pursuing our next generation hardware with a larger linear address space.

One other IAR banking overhead issue that I alluded to above but forgot to mention in yesterday's post is that each banked call pushes an effective 3-byte address on the stack instead of the normal 2, using up stack space that much quicker.

Michael (or anyone else that knows) -
Does the Keil banking method use up more stack space than a non-banked system?

Jeff