mmcp@sdsu.edu ( mmcp) (09/03/89)
Hi....
I have some questions about transputers which are unanswered because
either the reading material I have does not cover them, or I haven't been
able to try them out myself.
1) How many processes are able to run in the memory space of a
transputer. In other words, my view of a single transputer is
that there are a certain number of processes residing locally
in the memory space of the transputer itself, much the same
as is found in tradition multiprogramming architectures.
2) Given n transputers, how would you, say, place process 'x' on
transputer 1, process 'y' on transputer 2, .... etc? Do you
have control over where a process ultimately executes?
3) Can you ask questions about the state of a process, and do
certain actions based on this state. For example, could I
detect that a process is no longer executing on transputer 'x'
and schedule another process to execute there in its place?
4) Do transputers really only allow two levels of prioritization
at the hardware level? If not, any ways around this? The idea
that only two levels are available is somewhat limiting for
what I need to do.
Basically, the reason I ask is because I wish to implement something
like ConcurrentC hosted on a transputer, and I want the ability to change
my processes priorites dynamically, and if a transputer processor becomes
available, I would like to detect this and migrate a process from a more
burdened processor to execute on the newly available processor. Any detailed
tutorials which give a good feeling for exactly what sort of control one has
over the processors and what processes they are running, including how to
change priorities, etc., would be appreciated ( via e-mail ).
Thanks,
Syd Logan
Grad Student in C.S.
San Diego State University
--
mmcp
Syd Logan
UUCP ...!ucsd!sdsu!mmcp
ARPA mmcp%sdsu.uucp@ucsd.edu