Hello all,
I'm implementing a parallel algorithm on a master-slave module(based on Image Processing example under the directory of mcsdk2.x.x..), and I devided my data into 2 part2 running on 2 cores. it works fine. but when I devided my data into 8 parts and running on 8 cores, the data I received get little bit distortion. I tested my algorithm on 1 core by simulating a parallel structure that storing data independently on different register and afterwards combine them. so the algorithm should be fine.
So my question is there might be problem with memory arrangement, for example #pragma DATA_ALIGN(Databuffer ,128);
Hope someone can responde me.
Thanks in advance.
Best regards,
Wendy
