> The problem may be more one of latency than of raw CPU power. If the
> DSP calculations only require a relatively few cycles but require them
> nearly immediately and at frequent intervals, this would boost the CPU
I'd be surprised if you could not get away with intervals/latencies in
the ms range. This isn't a problem, as the scheduler has a frequency
in the same region.
> requirements. Add in the delay caused by, e.g., a disk driver that
> leaves interrupts masked for a considerable length of time, and the
> danger of sync loss increases.
This is the only real problem I've had with my software radio modem
drivers over the last years.
> I'd expect the price of DSPs and related modem parts to continue to
> drop, if not so quickly as Pentiums. Once the price of a DSP drops
Not as quickly as Pentiums? If you look at the sheer volume of DSP
chips sold by TI, ADI, Mot and others, I'm pretty sure the economies
of scale kick in much faster than for Intel. For example, every harddisk
these days uses at least one DSP (core), and my PC has about 3
harddisks,
but only one (non Intel :-)) CPU.
Tom
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html