Re: 2us Interrupt latency's for Linux 2.0.xx

Manoj Apte (manoj@ERC.MsState.Edu)
Wed, 14 Oct 1998 16:43:51 -0500 (CDT)


Sir,

I am assuming, this mail was sent out to the linux-kernel newsgroup by
mistake. But, since it was posted here, I thought it would be
but appropriate to post a reply to the same, in case others were
interested in knowing more.

Dr. Yodaiken,

I see that you are quite misinformed about our work at MSU. Our
project involves development of a Realtime messsage passing standard.
(MPI/RT) (ref: www.mpirt.org). We are in process of building the first
prototype of MPI/RT middleware and realtime network of workstations as a
technology demonstrator.

MPI/RT poses several middle-out constraints on the underlying operating
systems, that none of the OS's available to us support. Hence our goals
require us to modify some existing operating system, to best fit our
needs. After looking at several OSs we decided that RT-Linux would be most
readily modified, since its source is freely available, it has been fairly
well documented, and it already has a large and growing user base. The
support for RT-linux has also been excellent over the mailing list.

RT-linux as it existed (version 0.5) was quite far from our final
requirements, so we have been working on modifying the kernel to adapt it
to support an optimal mpirt implementation. For instance, we developed our
own time based scheduling mechanisms using the APIC that rtlinux is
incorporating only now. We also have been working on realtime drivers for
highspeed networks. In this, many of the ideas from the MARUTI OS are
being used.

We are not reimplementing anything, mostly adapting things to our
needs. It is unfortunate that you were led to believe the contrary, and
caused you such concern. I certainly hope this mail would help resolve any
misconceptions.

Manoj Apte
HPCL Lab
Mississippi State University

On Sun, 11 Oct 1998 yodaiken@chelm.cs.nmt.edu wrote:

> On Wed, Sep 02, 1998 at 09:27:31PM -0500, Manoj Apte wrote:
> >
> > I am using the pentium tsc to get the timings.. I am sending my algorithm
> > in the next email.. so you can look at it
> > and yes the delay for the 8253 is about 12us
>
> Since you folks got a big DARPA grant for reimplementing my work, you can
> hire me or Ingo to consult on this problem.
>
>

-
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/