[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [sc-dev] linux



Ronald J. Kuivila wrote:
>   In doing the sample accurate scheduling, I notice that on MacOSX there
> is about 0.05 second latency (the default value in Event) between sending
> a command to the server and having it executed.  How does this compare to
> Linux?

i've yet to proceed with an sc-client-application to come up with
definite measurements. how do you measure latency from within
supercollider?

i'm not sure what you mean by sample accuracy in event dispatching;
events are quantized to at least the clock resolution, no?

>   If the latency remmains consistent, it is probably not a major problem.
> But, it would be interesting to know if it were possible to run the server
> and client in the same process (but different mach threads) and use some
> kind of faster communication scheme.  (It would also be nice if the client
> and server could share audio buffers for oscilloscope style displays.)

i've thought about this too, and i guess it's quite doable: you could
write commands to the message fifos directly, without the socket
processing overhead. in this scheme, the current network layer would
just be another in-process client.

another approach for data sharing, the one that's partly implemented, is
shared memory; you should be able to access the server's buffers from
any client on the same host (not working on linux yet).

<sk>