Received: from sun2.nsfnet-relay.ac.uk by karazm.math.UH.EDU with SMTP id AA02470 (5.65c/IDA-1.4.4 for ); Thu, 24 Oct 1991 13:18:32 -0500 Received: from computer-science.manchester.ac.uk by sun2.nsfnet-relay.ac.uk via JANET with NIFTP id <13545-6@sun2.nsfnet-relay.ac.uk>; Thu, 24 Oct 1991 13:40:39 +0100 Date: Thu, 24 Oct 91 13:00:59 BST From: Dave Snowdon Message-Id: <9110241200.AA23286@r2i.cs.man.ac.uk> To: glove-list@KARAZM.MATH.UH.edu Subject: Transputers Here at Manchester we have a small vr group and were experimenting with some existing transputer hardware built here at Manchester. One thing to be wary of is that the quoted link speed (for a T800) is not actually acheivable. For every 8 bits of data the link engine actually sends 11 bits. Also there is a 2 bit acknowledgement, so if you're trying to get bi-directional communication the link engine is effectively sending 13 bits for every useful 8 bits of data. This gives a max transmission rate of something less than one megabyte/sec. Because of overheads of setting up a communication you also need to send messages of several bytes to approach even this. Some other hardware we're experimenting with is a frame-buffer which exists as shared memory between 4 transputers. We can then use 16 links to send data to the frame buffer. Dave Dave Snowdon (snowdond@cs.man.ac.uk) The Devil may care... But I don't mind (T.S.o.M.)