[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Priority revisited: a new primitive

Gerald Hilderink wrote:
> In my response about the PAR PRI in my previous message, I mentioned that
> the PRI PAR could be used in a hardware compilation. I would like to make a
> correction, because I think that a PRI PAR in a hardware compilation is a
> PAR. In a hardware compilation there is no shared resource on which priority
> can be used as a scheduling parameter -- there is no scheduling. So, PRI
> before PAR has no meaning in hardware. For the PRI ALT this is different.
> The PRI ALT has a shared resource on which priority is used as a scheduling
> parameter, namely the choice mechanism. This applies for hardware and
> software.

I will get back on this thread later, I hope. But one is often very
short of resources in hardware compilation, even with today's chips.
Think of an FPU, for example. So even in hardware compilation we may
need to share resources.
Whether one would wish to use something like PAR PRI to do that is
another matter.

>From the various responses to my original message, it is clear the I
have completely failed to explain the semantics of PAR PRI. I have
emailed some of you privately with what I hope is a slightly better

Too busy to reply properly just now.

Dr A E Lawrence (from home)