Subject: RE: [vhdl-200x] RE: Posix Threads
From: Scott Thibault (thibault@gmvhdl.com)
Date: Thu Jun 12 2003 - 05:12:14 PDT
>
>
> Jonas Nilsson wrote:
> >
> > Peter Ashenden wrote:
> >
> >>I'd just like to reinforce the view that VHDL already has
> threads - they're
> >>called processes. The issue is that they're statically created
> and there is
> >>no form of abstraction (ie, no declaration and instantiation).
> An proposal
> >>to add dynamic thread should build on the existing concurrency
> model in the
> >>language so as to main conceptual consistency. Hence the
> approach we took
> >>in SUAVE - see www.ashenden.com.au/suave.html.
> >
> >
> >
> > I second Peter's opinion.
> > The thing we need is a simple and clean way to dynamically
> create and destroy
> > processes. Not a completely new process/thread/whatchumacallit paradigm.
>
> I've had a quick look at the tutorial and I like the dynamic process
> stuff in SUAVE, but how much use is this to the verification people? It
> does seem very hardware-centric. Some comments/questions:
I think it is too hardware-centric. Since the goal is not to model actual
hardware the generic and port map stuff seems unnecessary and heavy.
> 2 How about concurrent procedures as well? This would save typing in the
> process and the wait bits, and would presumably generalise the call
> interface. It's also a much more "software" point of view.
I like this idea of concurrent procedure invocation. It would achieve the
same result as a pthread_create call, but more naturally. I think it would
require only minor changes to the language. This approach also would allow
the dynamic threads to share drivers for free without more verbose syntax
additions.
--Scott Thibault
Green Mountain
Computing Systems, Inc.
http://www.gmvhdl.com
This archive was generated by hypermail 2b28 : Thu Jun 12 2003 - 05:14:54 PDT