agillesp@undeeg.UUCP (Andrew W M Gillespie) (04/09/90)
I am posting this letter on the net in the hope that someone can answer
a host of questions I have concerning Helios and the transputer. The
transputer-group of which I am a member is at present working on,
inter-alia, an application involving distributed process control of an
industrial plant. The intention is to write most of the software in
OCCAM although some use may be made of 'C'. Helios is to be used as a
high level interface and O/S, to both distribute and manage the
controlling tasks within the plant.
In the available literature on Helios only higher-level languages such
as C, FORTRAN or PASCAL are referred to as being used to write tasks
that run under Helios. Given Helios's uniform inter-task interface the
same should be true of OCCAM.
* Is it possible to use OCCAM 'tasks' running under Helios ? * If so,
how does one go about doing exactly this (ie. format of OCCAM
EXEs/PROGRAMS etc) ? How does one obtain standalone OCCAM
EXEs/PROGRAMs suitable for running under Helios ?
It is possible to withdraw the Helios nucleus from specified nodes ( ==>
'native') so that the application tasks running on these processors have
full access to the 'raw' processing power without system overheads.
* If this is done, can Helios still be used to load tasks down to
these 'native' processors ? (say each 'native' node has at least
one Helios node adjacent to it) Can data be communicated to/from
these 'native' nodes by the adjacent Helios node ? * Can the Helios
nucleus be withdrawn/re-inserted dynamically (is this necessary) ?
Supplied with Helios are two reset-codes, namely 'im_ra_b4.d' and
'pa_ra.d' and another 'tram_ra.d' is supplied with the first update
disk. The default resource-map (RM) for a single B004 board uses
'im_ra_b4.d', but when I rewrote the RM to include an external B003
board, the latter reset-code did not appear to work, giving a message
'unable to boot network'. Reflecting on this, it seems as though this
code indicates that each node must be individually reset and then booted
from a neighbour. This would then obviously not work on the B003 which
has all the reset lines tied together. If the RM only included the
first node on the B003 as well as the b004, the network appeared to
reset and boot fine with im_ra_b4.d. I then tried the code 'tram_ra.d'
as 'pa_ra.d' did not work either. With 'tram_ra.d' the network appeared
to reset and boot correctly but then once logged-in, programs such as
the C-compiler 'cc' and the editor 'emacs' would not function; 'cc'
returning 'c failed because /helios/bin/cc returned with return code of
65280' and 'emacs' returned 'exec format error'.
* What reset-code should be used with the B004/B003 combination ? *
What are the generalised requirements of a reset-code, and is it
possible for the user to write reset programs for different types of
networks ? * Is code available for those reset 'codes' supplied with
Helios ?
I would be most grateful for any help that I can get. Thanks
Andy Gillespie
--
Andrew Gillespie, Dept. of Electronic Eng., Univ. of Natal, Durban, South Africa
Internet: agillesp.UNDEED@f4.n494.z5.fidonet.org UniNet-RSA: agillesp@UNDEED