On Mon, 2013-06-03 at 20:33 -0700, Jon Taylor wrote:
> 
> 
> 
> On Mon, Jun 3, 2013 at 5:35 PM, David Koontz <[email protected]>
> wrote:
>         
>         On 4 Jun 2013, at 6:13 AM, Jon Taylor <[email protected]>
>         wrote:
>         
>         > Setting stop-delta to 2 billion seems to make the simulation
>         > run long enough to enable me to test my code - the native
>         > sockets I'm using to emulate a bus aren't timing out any
>         > more. 

I would echo and reinforce David's opinion that increasing or removing
the delta limit per simulation timestep is almost certainly the wrong
approach, as well as restricting people to using GHDL only. 

If your process are self-timed, it is probably more appropriate to
approximate reality by adding some sense of timing to them; for example
by adding after clauses - even "after 1 ps" to your synchronisation
signal updates (all other updates can remain "immediate"). This would
reset the delta cycle count and allow the delta cycle mechanism to catch
and help identify any unexpected loops.

> I actually would prefer to use a separately-threaded back-end for
> general-purpose I/O and bus management (interrupts, exceptions,
> mcopy(), mmap(), etc) when running on a CPU-based system. 

It sounds to me that Ada 2005 or 2012 would be a good match for the
software levels of your design, with its close compatibility to VHDL
helping ease the transition between hardware and software levels. You
may not need the full extent of its multitasking capabilities, a
relatively lightweight subset like Ravenscar might be all you need.

- Brian



_______________________________________________
Ghdl-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/ghdl-discuss

Reply via email to