Hi everyone. I’m trying to use the cusparseScsrmm function to perform a multiply between a nn square sparse matrix and a nm tall matrix. However the function keeps giving error 3 in the beginning. Then I went through the sample program on the library and found that it set the ldc parameter to n+1 instead of n. I did it and it worked. However the result had an additional zero in every row, which would be very annoying. Any ideas how I can get rid of them? Either make the function work properly or removing the zeros with some simple actions (without uploading the data to host) would be good for me.
Thank you!