Conversation
| m_fb = factory::fb_factory<sopt::algorithm::ImagingForwardBackward<t_complex>>( | ||
| factory::algo_distribution::mpi_serial, m_measurements_distribute_grid, wavelets, m_uv_data, | ||
| m_sigma, beta, gamma, m_imsizey, m_imsizex, m_sara.size(), state.range(3), true, true, false, | ||
| 1e-3, 1e-2, 50, tf_model_path, nondiff_func_type::Denoiser); |
There was a problem hiding this comment.
Is the measurement operator normalisation calculated anywhere? It doesn't appear to be in AlgoFixtureMPI.
There was a problem hiding this comment.
No? I'd assume it just uses the default 1.0
mmcleod89
left a comment
mmcleod89
left a comment
There was a problem hiding this comment.
I think it's fine as long as we can confirm that the measurement operator is being normalised properly (if that's necessary for this benchmark), and if we don't need the code in the if(false) statement then we should probably just delete it.
|
Something is broken with homebrew on the MacOS runners 😞 Today it works 🥳 |
kmulderdas
left a comment
kmulderdas
left a comment
There was a problem hiding this comment.
Other than a parameter change in one of the benchmarks, I don't see any problems
kmulderdas
left a comment
kmulderdas
left a comment
There was a problem hiding this comment.
Addresses the changes I requested.
|
Ubuntu CI is failing due to actions/runner-images#11926 |
3 participants
No description provided.