Is there any quick start guide for programmers to write DSP accelerated applications for the TMS320C64x?
I have a program with a custom algorithm (not fft or usial filtering), and I want to speed it up using a multiprocessor coprocessor. So, how do I change the source to move the calculations from the main processor to the DSP? What are the limitations for DSP code?
I have experience with CUDA. In CUDA, I have to mark each function as a host, device, or entry point for a device (core). There are also functions to launch kernels and upload / download data to / from the GPU. There are also some device code restrictions described in the CUDA Reference Guide. Hopefully there is a similar interface and documentation for DSP.
source
share