BARKER@v1.ph.rhbnc.ac.uk (08/24/89)
Hello all,
I have this problem that seems to be totally ignored in the Transputer
Databook (1st ed. 1989). Can a T414 (or any other transputer for that matter)
maintain simultaneous bidirectional communication along a link?
Specifically:
I have an application where a T414B is acting as a host for a network of
T800s. In my original version of the host's program, the T414 would send out a
data packet (typically 10 - 1000 bytes), and wait to receive the results (again
10 - 1000 bytes). One these had been received, the host would send out the next
data packet, and so on...
I then decided that the program would be much more efficient if I had two
processes running in parallel on the T414: one sending data and one receiving
results. As soon as the receiver process received the first byte of the
results, it would pass a message to the sender. The sender would then start
transmitting the next data packet at the same time as the receiver was
receiving the rest of the last set of results.
In occam, very roughly:
PAR
WHILE TRUE -- receiver
SEQ
from.network ? results[0]
receiver.to.sender ! signal
from.network ? [results FROM 1 FOR results.length-1]
WHILE TRUE -- sender
SEQ
receiver.to.sender ? signal
to.network ! [data FROM 0 FOR data.length]
However I found no improvement in program speed. Now, this could be because of
some other bottleneck in my program, or simply because the T414 cannot maintain
the communication in both directions at the same time. Does anybody know the
answer?
--------------------------------------------------------------------------------
Simon Barker Phone: (0784) 439900 (UK)
Department of Physics, (+44) 784 439900 (international)
Royal Holloway and Bedford New College,
Egham Hill, Egham, JANET: UHAP033@UK.AC.RHBNC.VAXA or
Surrey, TW20 0EX, England BARKER@UK.AC.RHBNC.PH.V1BARKER@v1.ph.rhbnc.ac.uk (09/04/89)
Thanks to all who replied to my earlier question about performing
bidirectional simulteneous communication on a transputer. The general consensus
of opinion was that a T414 can send and receive at the same time, though at a
slightly reduced efficiency. Therefore there must be a problem somewhere else
in my program. I'll have to look into it.
Simon
--------------------------------------------------------------------------------
Simon Barker Phone: (0784) 439900 (UK)
Department of Physics, (+44) 784 439900 (international)
Royal Holloway and Bedford New College,
Egham Hill, Egham, JANET: UHAP033@UK.AC.RHBNC.VAXA (*) or
Surrey, TW20 0EX, England BARKER@UK.AC.RHBNC.PH.V1
(* This one may be unreliable for the next few weeks)