Add GPU offload capable sparse matrix#241

Closed

l90lpa wants to merge 3 commits intoecmwf:developfrom JCSDA-internal:feature/add-gpu-offload-capable-sparse-matrix

+600-22

Commits on Nov 25, 2024

Add SparseMatrix class into Atlas with host-device memory management.

l90lpa
authored and
wdeconinck
committed
Use atlas::linalg::SparseMatrix class instead of eckit::linalg::SparseMatrix.

l90lpa
authored and
wdeconinck
committed
Add hicsparse matrix_multiply to test_linalg_sparse_matrix_gpu
wdeconinck
committed