BTW, in re: cycles needed for the winmodem drivers --- somehow, the term
"real-time" keeps occurring to me. Certainly it can be done, but will we
end up slowing down the rest of the system?
The other question to keep in mind is latency; for any application where
interactive performance is an issue, or where the protocol isn't smart
enough to do windowing, the latency of the modem can affect your overall
performance just as much (if not more) than the actual speed (read:
bandwidth) of the modem.
It would seem to me that if you're forcing the CPU to do the DSP in
software, there's a very good chance that latency would be worst than if
you had a the $3-5 dollar hardware DSP in the modem.
Also consider that the $30 price differential may very well make a
performance difference on your system --- not just in terms of CPU load
(and increased number of interrupts, etc.), but also in the performance
of your modem on your system. This might not be as noticeable on a
Windows machine doesn't efficiently use the CPU anyway, but the effect
would almost certainly be more noticeable on a Linux machine.
Personally, I usually buy a ~$200 (high-end) external modem myself,
anyway, because I don't trust the construction and quality on the
sub-$100 internal modems, Winmodems or not. When everybody was
complaining about all of the glitches and problems with the USR
Sporsters, I was using my USR Courier modem with nary a problem.
Coincidence? I think not. You get what you pay for.
- Ted
-
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