Add benchmarks for special matrices
YingboMa opened this issue · comments
Aside from the differences of BLAS and LAPACK in Julia and Matlab, it maybe meaningful to compare the algorithms between Julia and Matlab. For instance, sqrtm
function requires to do a Schur decomposition for a normal square matrix which highly depends on LAPACK. How about benchmark sqrtm
for an upper triangular matrix?
I will soon have MATLAB R2017a.
I intend to re run the benchmark then.
I will add your suggestion as a new test.