Commit cc27732b authored by bliven_s's avatar bliven_s

Merge branch 'cryocare' into 'master'

Update cryocare environment to work on GPU

See merge request Pmodules/buildblocks!83
parents b4fbe6b5 17c57105
......@@ -13,8 +13,10 @@ Now configure jupyter
## Package notes
- cudatoolkit version is fixed because it needs to match merlin6's CUDA module
(required by the MotionCor2 package).
- This environment installs packages for gpu support, but these currently don't
seem to be working on merlin.
- cudatoolkit version should ideally match merlin6's CUDA module
(required by the MotionCor2 package). However, tensorflow needs cuda 10.0 while
no MotionCor2 binary is available. For now, having a different version in
conda and modules is working OK, but it could cause problems with future compilation.
- You can test GPU support by running this command on a GPU machine:
`python -c 'import tensorflow as tf; print(tf.test.is_gpu_available())'`
......@@ -4,30 +4,23 @@ channels:
- defaults
dependencies:
- python=3.7
- dask
- dask-jobqueue
- deap
- docopt
- graphviz
- ipython
- ipywidgets
# Explicit cryocare dependencies from singularity
- tensorflow-gpu=1.15.0
- keras
- matplotlib
- tifffile
- mrcfile
- nb_conda_kernels
- numpy
# Other dev packages
- jupyter
- dask
- dask-jobqueue
- pandas
- plotly
- pytables
- python-graphviz
- scikit-learn
- scipy
- seaborn
- tensorflow=1.13.1
- tensorflow-gpu
- tifffile
- cudatoolkit=9.2
- cudnn
- plotly
- pytables
- pip
- pip:
- csbdeep
- pixiedust
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment