Part Number: TMS320C6678 Tool/software: Code Composer Studio I found the DSP of my TCI6614 hardware emulation board is quite slow, so I have done some cycle measurement on the simulator( TCI6616 Device Cycle Approximate Simulator ) of CCS5. I think it is very close for these two DSPs. The result shows that it costs almost 1.9M cycles for 100000 times of addition, and approximately 19 cycle for each add calculation . Below is the simple test program: int sum = 0; int i = 0; for(i = 0; i < 100000; i++) { sum = sum + i; } This surprises me, and I do not understand it because my knowledge is that one cycle can have multiple additions be done, which benefits from multiple calculating units in C66xx DSP. I guess it is probably because I compiled the program with latest compiler(cl6x v8.3.2). Then I switch to v7.4.4, and the result is the same. I also test the program on TCI6614 emulator, and the result vary from different memory configurations. At most, it can reach 24 cycles per addition for HW (a little slower or worse than simulator). Could you please tell me what's wrong with my tested C66xx DSP? Is it supposed to be much faster? Thanks.
↧