Re: a joint letter on low latency and Linux

Bill Huey (billh@gnuppy.monkey.org)
Thu, 29 Jun 2000 19:45:07 -0700 (PDT)


Cory,

> Can you characterize what you mean by extensive? You can do FP in
> RTLinux. If you're going to be doing heavy computation you're not going to
> want to hold the cpu in RTLinux and should ship the data off to a
> user-level task to do the number crunching. Do you need to respond to an

Pushing it to as high a CPU useage as possible. ;-) The more a CPU can be
dedicated to doing signal processing the better. And I agree that this stuff
should be in user space.

> event, in realtime, and provide a response (that requires FP) within a
> given interval?

That an other evil stuff that's best kept in a single space, preferabley user
space, so that all of it and can be modified on the fly. Adjustment of DSP modules,
various kinds of timing, fairly complicated event models, etc...

> It does do SMP. In fact, we've had good numbers from kicking Linux off a
> processor and doing just RTLinux on it. Can you characterize what exactly
> you need RTLinux to do with regard to SMP for you? If so, we can tell you
> if it can or can't do it :)

I haven't been keeping track of this feature of RTLinux and I assumed that
little has change since a year or two ago.

bill

-
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.tux.org/lkml/