Closed chrisrichardson closed 1 year ago
tabulate_tensor_void
is an option to benchmark the cost of dolfinx assembly loop.
Since we can now just pass kernels directly in dolfinx, is this necessary?
Not really. Also, arguably, benchmarking dolfinx assembly loop is only for advanced users if ever needed, and they could pass void kernels themselves.
Remove unused options:
tabulate_tensor_void
,padlen
,assume_aligned
.