Im using the memcyp function to transfer vectors around, both internally and externally. When I transfer 1024 floats from internal memory to SDRAM it takes 2640 processor cycles but if i transfer the same floats from SDRAM to internal memory it takes 20656 cycles. I know the SDRAM is significantly slower and requires clock alignment but.. a factor of ten difference?
Any comments ?
PS.. does anyone have DMA code to transfer from internal memory to SDRAM and vice versa for 21369 processor ?