Posts: 384
Threads: 18
Joined: Sep 2010
Quote:
Is 128 kB enough?
For an AVR target a KEMU config including 11c/12c/15c/16c support
weighs in at:
/usr/bin/avr-size kemu_m1284p.elf
text data bss dec hex filename
51498 162 2239 53899 d28b kemu_m1284p.elf
Note this in on a modest 8-bit AVR core which isn't exactly
the runtime architecture Dennis Ritchie had in mind. So I'd
expect ARM thumb code to do considerably better. Note however
the firmware occupies the lion's share of flash, thus any
reduction due to use of a sane runtime architecture will be
proportional to actual emulator code.
Even a 64KB sam7l would suffice if the above assumption holds.
Interestingly 24-28KB of flash is more than sufficient for a
single 15c configuration, and a 12c flash footprint would
further reduce that by 7.5KB. I'd expect quite a bit of elbow
room could be found in the sam7l128 used in the encore 12c/15c.
RWM (bss + data above) however would need to be partitioned
between the 2KB of battery backed and 4KB volatile regions.
But I think that is doable. Other features such as a CLI
interface and support (mapping the segment display to ascii,
etc..) are included in the above image, the RWM storage
for which doesn't need to be persistent over power off
periods or can be deconfigured altogether.
Posts: 255
Threads: 22
Joined: May 2011
I am going to take a wild guess and say that though they look very similar, that the 12c+ and the 15c LE are actually a little bit different.
The 20b and the 30b have some minor difference that allows the 30b F/W to recognize that it is running on a 20b chassis. On the 20b chassis, HP was a bit clever here and the 30b F/W runs in a reduced "20b" mode -- none of the 30b-specific functions run.
So, given that I own neither a 12c nor a 15c LE, that if you could get ahold of the 15c LE F/W, that it would not run on the 12c chassis.
Note that this is likely a one direction limitation. We have seen that 12c F/W does run on a 15c LE just as the 30b will run the 20b F/W.
Just a guess...