RE: [linrad] Linrad user interface

["Leif Åsbrink" <SMTP:LEIF.ASBRINK@xxxxxxxxxxxxxxxxxx>]

Hi Edson,
  
> One possible way to free the user interface dependency on one single API
> is to separate the user interface code from the DSP code completely
> (different processes). The communication could be done using Internet
> sockets and a simple client-server protocol. With this approach, user
> interface developpers need only to follow the protocol specifications and
> DSP developpers don't have to worry about user interface issues. Another 
> advantage is that the DSP code could even run on a separate machine.
> 
> What do you think, Leif?

I am not a programmer, I have no idea about the consequences of this
proposal. Linrad has many dsp processes and each one has its own 
graphical routines. Some of the processes feed information back to
earlier processes.

Do you know what the cpu load might be when sending 12 megabits/s
using Internet sockets ? The delay has to be less than 0.02 seconds
each time. To be a realistic option for the next few years it must
not take more than 5% of the cpu time on a 500 MHz PentiumIII to
have two such links if all is done on the same computer.

If the processing were split on several computers it would be a 
lot easier, but then it will be very difficult to make a good
receiver of the entire system because there would be many
computers to put into RFI tight boxes.

73

Leif



LINRADDARNIL