[evla-sw-discuss] RTOSes ?

[B Carlson]

I think that a one size fits all RTOS is a desirable goal, but I don't
think it is realistic given the constraints of the antenna MIBs, and
what I see as a full-featured RTOS requirement for the correlator MIBs.
I'm really attracted to "Linux everywhere", but Bill raises some serious
concerns that simply can't be ignored.  For the correlator, I want to
get a MIB evaluation kit (which for now means the CompuLab PC/104+
board), and try Linux out for size to see what happens.  I've heard some
talk that Linux is being slagged because its in competition with
Windows--I'd just like to see for myself to be sure.  However, I'm fully
prepared to use VxWorks if there are any doubts about RT Linux.  The
CompuLab docs say that it has ready to run packages for Linux, Windows
CE, Windows 95/98/2000/NT, VxWorks, and DOS.  I think only VxWorks is a
viable back-up plan to Linux.

Would VxWorks for the custom-designed antenna MIB be possible in terms
of design effort for the BSP, size of OS, and cost?  I don't know.  I
would not like to have take up design effort getting some other RTOS to
work on the correlator MIB, just so it is compatible with the antenna
MIB.  So VxWorks *may* be the only chance for a one-RTOS system.  I
think that Linux could very well be used for the quasi-real-time
computers (higher-level computers, correlator backend etc), contrary to
Bill's opinion.  We have had very good success here running a Beowulf
cluster with Linux.

Brent.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Brent.Carlson.vcf
Type: text/x-vcard
Size: 363 bytes
Desc: Card for B Carlson
URL: <http://listmgr.nrao.edu/pipermail/evla-sw-discuss/attachments/20020129/4dcdd784/attachment.vcf>