Tool/software: Code Composer Studio
Hi TI,
i use the code "psdk_rtos_auto_j7_06_02_00_21/tiadalg/tiadalg_image_preprocessing/alg/tiadalg_image_preprocessing_c66.c"
i debug this at C66 core with CCS
in fact, this code uses _mem8 , _loll and _hill intrinsics
and each instruction will cost more than 600 cycles
for example, "*dst_ptr_r_u8++ = (uint8_t)(_loll(u64temp1) >> 24);" will cost 600+ cycles
i think it is very slow and cost too much cycles
Does this caused by prefetch disabled?
if so , can you tell me how to enable the prefetch
can you give me any suggestions?
thanks very much
