Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add caffe #256

Merged
merged 1 commit into from
Apr 5, 2016
Merged

Add caffe #256

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 20 additions & 0 deletions recipes/caffe/build.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
#!/bin/bash

# Setup CMake build location
mkdir build
cd build

# Configure, build, test, and install.
cmake -DCPU_ONLY=1 -DBLAS="open" -DCMAKE_INSTALL_PREFIX="${PREFIX}" ..
make
make runtest
make install

# Python installation is non-standard. So, we're fixing it.
mv "${PREFIX}/python/caffe" "${SP_DIR}/"
for FILENAME in $( cd "${PREFIX}/python/" && find . -name "*.py" | sed 's|./||' );
do
chmod +x "${PREFIX}/python/${FILENAME}"
cp "${PREFIX}/python/${FILENAME}" "${PREFIX}/bin/${FILENAME//.py}"
done
rm -rf "${PREFIX}/python/"
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is pretty hacky here. However, the install just dumps all of the Python stuff in ${PREFIX}/python, which is not cool. So, we try to resort it based on where it should have gone. Should work with upstream to fix this.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Point them to cookiecutter: https://github.com/audreyr/cookiecutter

Shouldn't need much work to use that for the drudge work, then just copy their whole folder structure in.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sure, will keep that in mind. Thanks.

121 changes: 121 additions & 0 deletions recipes/caffe/meta.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,121 @@
{% set base_version = "1.0.0" %}
{% set version_postfix = "rc3" %}
{% set version = base_version + version_postfix %}

package:
name: caffe
version: {{ version }}

source:
fn: caffe_{{ version }}.tar.gz
url: https://github.com/BVLC/caffe/archive/{{ version_postfix }}.tar.gz
sha1: 79e075d06bea856164c926b92a40242b0369965b

build:
number: 0
skip: true # [not (linux and py27)]
features:
- nomkl

requirements:
build:
- cmake
- cython
- nomkl
- openblas
- boost
- hdf5
- gflags
- glog
- h5py
- hdf5
- ipython
- leveldb
- lmdb
- matplotlib
- networkx
- nose
- numpy
- opencv 2.*
- pandas
- pillow
- protobuf
- python
- python-dateutil
- python-gflags
- python-leveldb
- pyyaml
- scikit-image
- scipy
- six
- snappy

run:
- nomkl
- openblas
- boost
- hdf5
- gflags
- glog
- h5py
- hdf5
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

hdf5 is an implicit dependency, I think. the h5py dependency will get it. Only include it if they drectly use the HDF5 API and link against it.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I believe they do link against it.

- ipython
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is IPython actually required for things to run?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Honestly, with most of the python dependencies I am not sure to what extent they are needed. Do they actually run something with them or are they simply installed for the user's benefit? Would be nice to get clarification/confirmation on this from upstream. At present, installing the kitchen sink seems like the safe way forward.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also, see this comment for more details.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Agreed. Pick your battles. This one makes me too eager to go in and make sure they have clean interfaces on each of the C++/python sides. In my infinite spare time, of course.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This one makes me too eager to go in and make sure they have clean interfaces on each of the C++/python sides. In my infinite spare time, of course.

Right?! There are a few worrisome things here for sure, but yes time is an issue. Hoping when they hit rc4 things will be a bit more sane.

- leveldb
- lmdb
- matplotlib
- networkx
- numpy
- opencv 2.*
- pandas
- pillow
- protobuf
- python
- python-dateutil
- python-gflags
- python-leveldb
- pyyaml
- scikit-image
- scipy
- six
- snappy
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's unclear how many of these are needed and how many are simply recommended for examples or similar. In any event, it is easier to bundle everything they specify for the system and for their python environment. Though a few we know to drop from run time like Cython.


test:
commands:
# Test commands.
- command -v "${PREFIX}/bin/caffe"
- command -v "${PREFIX}/bin/classification"
- command -v "${PREFIX}/bin/classify"
- command -v "${PREFIX}/bin/compute_image_mean"
- command -v "${PREFIX}/bin/convert_cifar_data"
- command -v "${PREFIX}/bin/convert_imageset"
- command -v "${PREFIX}/bin/convert_mnist_data"
- command -v "${PREFIX}/bin/convert_mnist_siamese_data"
- command -v "${PREFIX}/bin/detect"
- command -v "${PREFIX}/bin/device_query"
- command -v "${PREFIX}/bin/draw_net"
- command -v "${PREFIX}/bin/extract_features"
- command -v "${PREFIX}/bin/finetune_net"
- command -v "${PREFIX}/bin/net_speed_benchmark"
- command -v "${PREFIX}/bin/test_net"
- command -v "${PREFIX}/bin/train_net"
- command -v "${PREFIX}/bin/upgrade_net_proto_binary"
- command -v "${PREFIX}/bin/upgrade_net_proto_text"
- command -v "${PREFIX}/bin/upgrade_solver_proto_text"
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could still add some Python binaries.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Agreed. It would be good to test at least one of the things you moved

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

They have been added.


# Test includes.
- test -d "${PREFIX}/include/caffe"

# Test libraries.
- test -f "${PREFIX}/lib/libcaffe.so"

imports:
- caffe

about:
home: http://caffe.berkeleyvision.org/
license: BSD 2-Clause
summary: A deep learning framework made with expression, speed, and modularity in mind.

extra:
recipe-maintainers:
- jakirkham