[Debian-pan-maintainers] nabu on grades

[Roland Mas]

I'm getting started on that :-)

Roland.

Le 21/11/2024 à 16:22, FARHI Emmanuel a écrit :
> Ouch !
>
> We do have pyvkfft and pycuda, but it seems we have missed the 
> scikit-cuda dependency for Nabu, which is indicated as optional in:
>
>   * https://gitlab.esrf.fr/tomotools/nabu/-/blob/master/pyproject.toml?ref_type=heads
>   * https://tomotools.gitlab-pages.esrf.fr/nabu/install.html
>
> Not so optional in fact...
> It may be the same story with the missing |pycudwt| for wavelets
>
> The good news is that the scikit-cuda source does not seem to require 
> too many hidden dependencies, see:
>
>   * https://github.com/lebedov/scikit-cuda
>
> Regarding the pycudwt, the pypi and source code are a:
>
>   * https://pypi.org/project/pycudwt/
>   * https://github.com/pierrepaleo/pypwt
>
> Roland, do you think you can push sk-cuda and |pycudwt| into NEW ?
>
> I hope there are no other hidden dependencies.
>
> Emmanuel.
>
> ------------------------------------------------------------------------
> *De: *"Andrew KING" <andrew.king at synchrotron-soleil.fr>
> *À: *"FARHI Emmanuel" <emmanuel.farhi at synchrotron-soleil.fr>
> *Cc: *"PIAULT Pierre" <pierre.piault at synchrotron-soleil.fr>, "PICCA 
> Frederic-Emmanuel" <frederic-emmanuel.picca at synchrotron-soleil.fr>
> *Envoyé: *Jeudi 21 Novembre 2024 15:19:19
> *Objet: *Re: nabu on grades
>
> Hi Emmanuel,
>
>   I have just tried Nabu again - I get another error relating to cuda 
> - see messages copied below...
>
> Can you have a look?
> Thanks, Andy
>
> king at grades-vm:~/persistent_area/benchmark$ nabu 
> 0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING/nabu.conf
> Browsing dataset
> dataset 
> /tmp/sshfs-persistent_area/benchmark/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING1.nx 
> has broken virtual-dataset at /flyscan/instrument/detector/data. 
> ('/tmp/sshfs-persistent_area/benchmark/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING/tmp/sshfs-persistent_area/benchmark/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING.nxs',) 
> missing
> Updating dataset information with user configuration
> Loaded darks from 
> /tmp/sshfs-persistent_area/benchmark/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING1_darks.hdf5
> Loaded flats from 
> /tmp/sshfs-persistent_area/benchmark/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING/0231_WF6_UJcani_LHS_T1_7p2_top_30k_400ms3600pr_tomo_BINNING1_flats.hdf5
> Overwriting pixel_size = 1.3e-06
> Doing dataset estimations
> Doing coupled validation
> Going to reconstruct slices (0, 1023)
> Unsharp mask margin: 21 pixels
> Estimated margin: 21 pixels
> Processing sub-volume ((0, 3600), (0, 1024), (0, 1024))
> Creating a new pipeline object
> Set sub-region to ((0, 3600), (0, 1024), (0, 1024))
> Traceback (most recent call last):
>   File "/usr/bin/nabu", line 8, in <module>
>     sys.exit(main())
>              ^^^^^^
>   File "/usr/lib/python3/dist-packages/nabu/app/reconstruct.py", line 
> 114, in main
>     R.reconstruct()
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/reconstruction.py", 
> line 621, in reconstruct
>     self._reconstruct_chunked()
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/reconstruction.py", 
> line 573, in _reconstruct_chunked
>     self._instantiate_pipeline_if_necessary(task, prev_task)
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/reconstruction.py", 
> line 551, in _instantiate_pipeline_if_necessary
>     self._instantiate_pipeline(current_task)
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/reconstruction.py", 
> line 543, in _instantiate_pipeline
>     pipeline = self._pipeline_cls(*args, logger=self.logger, 
> margin=task["margin"], **kwargs)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/chunked_cuda.py", 
> line 61, in __init__
>     super().__init__(
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/chunked.py", 
> line 106, in __init__
>     self._init_pipeline()
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/chunked.py", 
> line 276, in _init_pipeline
>     self._init_phase()
>   File "/usr/lib/python3/dist-packages/nabu/pipeline/utils.py", line 
> 20, in wrapper
>     return func(*args, **kwargs)
>            ^^^^^^^^^^^^^^^^^^^^^
>   File 
> "/usr/lib/python3/dist-packages/nabu/pipeline/fullfield/chunked.py", 
> line 481, in _init_phase
>     self.phase_retrieval = self.PaganinPhaseRetrievalClass(
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>   File "/usr/lib/python3/dist-packages/nabu/preproc/phase_cuda.py", 
> line 45, in __init__
>     self._init_fft(fft_backend)
>   File "/usr/lib/python3/dist-packages/nabu/preproc/phase_cuda.py", 
> line 66, in _init_fft
>     self.cufft = fft_cls(shape=self.data_padded.shape, 
> dtype=np.float32, r2c=True)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>   File "/usr/lib/python3/dist-packages/nabu/processing/fft_base.py", 
> line 46, in __init__
>     self._configure_batched_transform()
>   File "/usr/lib/python3/dist-packages/nabu/processing/fft_cuda.py", 
> line 48, in _configure_batched_transform
>     raise ImportError("Please install pycuda and scikit-cuda to use 
> the CUDA back-end")
> ImportError: Please install pycuda and scikit-cuda to use the CUDA 
> back-end
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/debian-pan-maintainers/attachments/20241123/6dcf9dcf/attachment.htm>