I have four questions: 1. For the dimensions of matrices B and C, are we assuming that the number of columns is fixed to be two? If not, would that mean a zero-element row in the original matrix would get a full row of zeroes in B and C (i.e. if I have 4 columns for matrix B and C, their zero-element row will have 4 consecutive zeroes?) 2. "...in the first case you can generate a random number between 1 and 100 for each entry of the matrix, and the entry is non-zero if the random number is between 0 and 1. Generate another random small integer between 1 and 10 and store it as the entry in that location of the matrix." I am a bit confused with this one, should the range of generated integers differ for each case? 3. Speaking of test cases, for each set, are we doing the matmul on the original matrices and their sparse versions for performance comparisons? 4. How should we approach finding the optimal number of threads?