Open Intron7 opened 7 months ago
With the new numba kernel for sparse scaling, we need to figure out how and when the numba code becomes faster than the array code. This might lead to numba completly replaceing the array operations for sparse matrices.
Would be a nice use case for working on the usability of benchmarks (related to #2977)
With the new numba kernel for sparse scaling, we need to figure out how and when the numba code becomes faster than the array code. This might lead to numba completly replaceing the array operations for sparse matrices.