I am having trouble with passing a symmetrix matrix with upper/lower stored in Numpy to Cublas Ssymm function.
I already changed the numpy array’s row major to Cublas’ column major.
That worked for Sgemm.
But for Ssymm, the result I get is partially wrong.
I printed out the array I passed in, and it was shown correctly, but after comuptation, the result seemed to be partially wrong.
So how do I pass in a upper/lower stored symmetric matrix from Numpy to Ssymm in Cublas?
Any help would be appraciated.
Never mind I just needed to transpose the matrix and it’s working now.