This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Latency: Interrupt Keyword vs. DSP/BIOS Dispatcher

All:

Setup: CCS 5.5, C6713, DSP/BIOS 5.42, CGT 7.4.6

Interrupt Keyword Measurement - 49 cycles

   I set up a breakpoint at the entry of the interrupt (in the interrupt service table) and a second breakpoint a the end of the Context-Save in the interrupt service routine. I then used the "profile clock" to make the measurement of 49 cycles.

 

DSP/BIOS Dispatcher Measurement - 307 cycles

   I set up a breakpoint at the entry of the interrupt associated with the dispatcher, and a second breakpoint at the beginning of the interrupt service routine (function). I again used the "profile clock" to make the measurement of 307 cycles.

 

As it turns out, the only DSP/BIOS api that is run in the second interrupt is PRD_tick( ). It looks like I can use the interrupt keyword instead of the dispatcher IF I make use of one of the internal timers (timer0 or timer1) to do the PRD clocking.

Another option appears to be using an assembly-base interrupt service routine and then using the HWI_enter and HWI_exit macros, allowing me to then call the PRD_tick( ) from the assembly routine. (A lot more effort!)

Are there other options that can allow me to do away with the dispatcher on the second interrupt?