Neurokernel is a Python framework for developing models of the fruit fly brain and executing them on multiple NVIDIA GPUs.
Neurokernel requires
- Linux (other operating systems may work, but have not been tested)
- Python 2.7 (Python 3.0 is not guaranteed to work)
- at least one NVIDIA GPU with Fermi architecture (or something more recent)
- NVIDIA's GPU drivers
- CUDA 5.0 or later.
To check what GPUs are in your system, you can use the inxi command available on most Linux distributions:
inxi -G
You can verify that the drivers are loaded as follows:
lsmod | grep nvidia
If no drivers are present, you may have to manually load them by running something like:
modprobe nvidia
as root.
The mpi branch of Neurokernel also requires OpenMPI 1.8.4 or later compiled with CUDA support. See this page for installation information. Note that OpenMPI 1.8 cannot run on Windows.
Some of Neurokernel's demos require either ffmpeg or libav installed to generate visualizations.
Download the latest Neurokernel code as follows:
git clone https://github.com/neurokernel/neurokernel.git
Since Neurokernel requires a fair number of additional Python packages to run, it is recommended that it either be installed in a virtualenv or conda environment. Follow the relevant instructions below.
See this page for virtualenv installation information.
Create a new virtualenv environment and install several required dependencies:
cd ~/ virtualenv NK ~/NK/bin/pip install numpy cython numexpr tables pycuda
If installation of PyCUDA fails because some of the CUDA development files or
libraries are not found, you may need to specify where they are explicitly. For
example, if CUDA is installed in /usr/local/cuda/, try installing PyCUDA
as follows:
CUDA_ROOT=/usr/local/cuda/ CFLAGS=-I${CUDA_ROOT}/include \
LDFLAGS=-L${CUDA_ROOT}/lib64 ~/NK/bin/pip install pycuda
Replace ${CUDA_ROOT}/lib with ${CUDA_ROOT}/lib64 if your system is
running 64-bit Linux. If you continue to encounter installation problems, see
the PyCUDA Wiki for more information.
Run the following to install the remaining Python package dependencies listed in setup.py:
cd ~/neurokernel ~/NK/bin/python setup.py develop
Note that conda packages are currently only available for 64-bit Ubuntu Linux 14.04. If you would like packages for another distribution, please submit a request to the Neurokernel developers.
First, install the libibverbs1, libnuma1, libpmi0, libslurm26, and
libtorque2 packages; these are required by the conda OpenMPI packages prepared
for Neurokernel. You also need to ensure that CUDA has been installed in
/usr/local/cuda.
Install conda by either installing Anaconda or Miniconda. Make sure that the following lines appear in your ~/.condarc file so that conda can find the packages required by Neurokernel:
channels: - https://conda.anaconda.org/neurokernel/channel/ubuntu1404 - defaults
Create a new conda environment containing the packages required by Neurokernel by running the following command:
conda create -n NK neurokernel_deps
Run the following to install the remaining Python package dependencies listed in
setup.py:
source activate NK cd ~/neurokernel python setup.py develop
To build Neurokerne's HTML documentation locally, you will need to install
- sphinx 1.2 or later.
- sphinx_rtd_theme 0.1.6 or later.
Once these are installed, run the following:
cd ~/neurokernel/docs make html
More information about Neurokernel can be obtained from the project website at https://neurokernel.github.io.
Neurokernel's documentation is available at http://neurokernel.readthedocs.org.
See the included AUTHORS file for more information.
This software is licensed under the BSD License. See the included LICENSE file for more information.
The Neurokernel Project is independent of the NeuroKernel Operating System developed by NeuroDNA Computer.