One way to use NICs would be to drop the protocol
stack and just use
the NICs as high speed serial interfaces. You could use cross-over
cables and point-point connections only, to completely avoid the risk
of collisions and the related "random" latencies that may cause.
right, i remember reading about a specialized yet simple-to-construct form
of network hardware developed at purdue university that allowed latency in
the microseconds for certain types of distributed apps. they use it to
cluster linux boxes though, not rt os's.
Naturally this
depends on a signal path separable into big chunks.
Yes... Or rather, the number of node->node hops is restricted by the
connection latency, internal latency of nodes and the required total
system latency. Use RTL or RTAI, fast NICs, cross-over point-point
connections and process say, 0.1 ms' worth of audio/MIDI data at a
time, and you can send data around quite a bit before you even reach
the "magical" 3 ms. Unfortunately, processing that small chunks of
data isn't terribly efficient, so at some point you'll undo the
advantage of having more CPU power available.
What i was thinking of only involves sending MIDI over the network, audio
produced by each machine would go directly to a mixing board and not back
into a computer. The point would be to perform musical pieces involving
many synth (or sampled) parts by breaking the number crunching down into
parallel sections simple enough to be handled by a single machine. For
example, if nine synths are playing concurrently, put 3 on each of 3
machines. Setup & teardown of the signal paths can happen in non-rt, but
the midi notes need rt delivery.
--jacob