Open alecandido opened 4 months ago
Current tests are closer to benchmarks, since they are testing the whole execution.
Individual functions should be tested as much as possible in isolation. This could make part of the tests more independent of CUDA.
As discussed during @scarrazza 's SG visit last week, unit tests can live in the pytest files for functionality validation, while benchmarking tests can live in qibojit-benchmarks repo.
pytest
qibojit-benchmarks
Current tests are closer to benchmarks, since they are testing the whole execution.
Individual functions should be tested as much as possible in isolation. This could make part of the tests more independent of CUDA.