Multiple Propellers
alex2
Posts: 10
I am trying to find a best way to efficiently communicate between 2 Propeller chips ( only 2 for now ).
My idea would be to use 2 I/O pins/lines on each chip to do this, in such a way that chip1 one would have Send Data Request/Received Data Acknowledged shared line/pin and second line/pin would be polling output from the other chip.
Chip2 would have the lines/pins in a reverse order. There is an issue though, when both chips
try to rquest data from each other at the same time; what exactly would happen is not clear to me.
So the issue I am having right now is how to handle SIMULTANEOUS requests for data from several processors ?
Does that make sense, any ideas ?
Somewhat simplistic for now but I am hoping that this will generate enthusiastic discussion of the subject.
Besides, has anyone worked on a project or idea that uses more than one Propeller yet ?
Thanx,
Alex
My idea would be to use 2 I/O pins/lines on each chip to do this, in such a way that chip1 one would have Send Data Request/Received Data Acknowledged shared line/pin and second line/pin would be polling output from the other chip.
Chip2 would have the lines/pins in a reverse order. There is an issue though, when both chips
try to rquest data from each other at the same time; what exactly would happen is not clear to me.
So the issue I am having right now is how to handle SIMULTANEOUS requests for data from several processors ?
Does that make sense, any ideas ?
Somewhat simplistic for now but I am hoping that this will generate enthusiastic discussion of the subject.
Besides, has anyone worked on a project or idea that uses more than one Propeller yet ?
Thanx,
Alex
Comments
"supercomputing" thread http://forums.parallax.com/forums/default.aspx?f=25&m=121344.
Yes , I am toying around for some time already with the idea of having a prop demoboard control a (number of) propeller(s) for tasks like video processing, or as a debug aid but have found not the perfect communication method either. The three prop example I posted is only a trial if I could get this to work.
My current thoughts are to use a master slave scheme, using one assembly coded Cog in the slave and the master, where the master continously polls the slave(s), sending one byte commands answered by a one byte slave response. Depending on the slave response then the master or slave initiates an N Byte data transfer. This scheme could be implemented for two props using only one pin, as switching the send/receive direction of the pin can be fully time determined.
Nico Hattink