I have to use asterisk with some dedicated DSP chips, which will do the
expensive G729 CODEC computing, so that the server processor has minimum
load. I was informed, I’ve to use GPAK to implement this. So far I’ve
researched, I’ve found that, Asterisk uses Dahdi to *talk* to external
Can you please direct me to references/documentations that clarifies,
1. The over all architecture of asterisk with dedicated DSP chips,
i.e., how does asterisk integrates with dedicated dsp chips?
2. Whether embedded scenario is a must for such development or is it
possible to use asterisk on server and dsp’s through some PCI/Ethernet?
3. What could possibly be the build process in such scenario? Tools
i.e., cross compilers etc?
4. How do asterisk know, that when a call arrives it has to use a
particular DSP chip for Encoding/Decoding?
The GPAK user’s manual states,
The physical layer of communication between the host processor and a G.PAK
> DSP is often dependent upon the hardware design. This interface can be via
> a host port interface (HPI), shared memory, a PCI bus, Ethernet or any
> other fast data pathway.
So far I’ve comprehended from the above quotation, it is possible to use
asterisk on a server and DSPs on some remote location (say, not built into
the same board where asterisk is running) . Am I correct?
Thank you very much for patient reading.