Re: [Simulavr-devel] connecting pins

[Michael Hennebry]

On Tue, 8 Jul 2008, Klaus Rudolph wrote:

>
> > If I find it, I'll still need to know how to connect tcl to simulavrxx.
> The example code is part of downloadable code. Please take a look for 
> anacomp! Here you see a tcl script running simulavrxx.
>
> > Also, I'm not so much interested in producing a pretty picture
> > as I am in debugging the code that produces that picture,
> > e.g. figuring out what causes that occasional flash
> > of red in an animation that is entirely blue and green.
>
> ?? If you not! want to see anything from the circuit, you have no need to 
> simulate it, if only outputs are required. So why you ask?

To be more precise, I should have written
that I'm not just interested in pretty pictures.
If nothing else, a pretty picture tells my I'm
doing something at least apprximately right.

I'd also like something that could "look" at said pictures
faster than I can and hit break when something is wrong.
Eyeballing a multimegabyte trace file isn't really an option.

> > Single-stepping won't cut it.
> > I might need to simulate several cpu-seconds
> > (9,600+ changes per seccond) before the simulated
> > hardware detects the error and hollers stop.
>
> No problem to simulate years of running if your hard disk is able to collect 
> the trace files :-)

And if the computer is fast enough.
A ring buffer would handle the storage issue
*if* the simulator knew when to stop.

> > Also, can I make Net or other connections
> > to the general purpose registers?
> ?? Connects are for physical pins with a logic level on them. What kind of 
> circuit should be connected to what kind of register?? Do you mean the CPU 
> Registers R0..R31? You have all outputs in the tracefile and it should be no 
> problem to have same also in any kind of debug output, if you modify the 
> avrcore.cpp file.

I want to be able to print or not or stop or
not based on some fairly complex criteria.
E.g. in message passing, the contents of a register are destructively
copied (indirectly of course) into a register of another processor.
It would be nice to be able to base debugging output on whether
message passing is in progress and whether it is successful.
This is roughly the printf method of debugging mathematical programs on PCs.
It's obviously not something that can be done on the actual hardware.

BTW did you mean avrdevice.cpp ?
There is no avrcore.cpp .

-- 
Michael   address@hidden
"Those parts of the system that you can hit with a hammer (not advised)
are called Hardware;  those program instructions that you can only
curse at are called Software."