Nnabla-ext-cuda

pip install nnabla
pip install nnabla_ext_cuda102 For CUDA 10.2 x cuDNN 8.0 users


For distributed training, You need to install the correct package that
matches the version of MPI installed on your system.

We prepared following packages.

- nnabla-ext-cuda100-nccl2-mpi2-1-1 (Ubuntu18.04 default)
- nnabla-ext-cuda100-nccl2-mpi3-1-6
- nnabla-ext-cuda102-nccl2-mpi2-1-1 (Ubuntu18.04 default)
- nnabla-ext-cuda102-nccl2-mpi3-1-6
- nnabla-ext-cuda110-nccl2-mpi2-1-1 (Ubuntu18.04 default)
- nnabla-ext-cuda110-nccl2-mpi3-1-6

If you want to use a version of MPI not listed above, you need to build it from the source.


Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

[Install Guide](http://nnabla.readthedocs.io/en/latest/python/installation.html)
To use C++ inference feature, follow the demonstration on MNIST inference in C++.

[Demonstration](https://github.com/sony/nnabla/tree/master/examples/cpp/mnist_runtime)

[Build Guide](http://nnabla.readthedocs.io/en/latest/python/build_on_linux_with_dt.html)
The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda110, nnabla-ext-cuda102, nnabla-ext-cuda100 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been (or going to be) closed.

+ nnabla-ubuntu16
+ nnabla-ubuntu18
+ nnabla-ext-cuda
+ nnabla-ext-cuda80
+ nnabla-ext-cuda91
+ nnabla-ext-cuda92
+ nnabla-ext-cuda101
+ nnabla-ext-cuda90-nccl2-ubuntu16
+ nnabla-ext-cuda100-nccl2-ubuntu16
+ nnabla-ext-cuda100-nccl2-ubuntu18
+ nnabla-ext-cuda101-nccl2-ubuntu16
+ nnabla-ext-cuda101-nccl2-ubuntu18

The following "nnabla-ext-cuda" docker images have been deprecated.
+ py27-cuda92
+ py36-cuda92
+ py37-cuda92
+ py27-cuda92-v1.0.xx
+ py36-cuda92-v1.0.xx
+ py37-cuda92-v1.0.xx

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

Python 2 is no longer be supported from v1.5.0.
CUDA8.0 is no longer be supported from v1.6.0.
Python3.5 is no longer be supported from v1.14.0.
CUDA9.0 is no longer be supported from v1.14.0.

pip install nnabla
pip install nnabla-ext-cuda101 For CUDA 10.1 x cuDNN 7.6 users


Additional setup may be required depending on your OS or environment. Please check Python Package Installation Guide for details.

[Install Guide](http://nnabla.readthedocs.io/en/latest/python/installation.html)

To use C++ inference feature, follow the demonstration on MNIST inference in C++.

[Demonstration](https://github.com/sony/nnabla/tree/master/examples/cpp/mnist_runtime)

For distributed training, you need to build a binary from source. See the guide for building multi-GPU training package.

[Build Guide](http://nnabla.readthedocs.io/en/latest/python/build_on_linux_with_dt.html)


The "nnabla-ext-cuda" package is temporarily unavailable. Use of this package is not recommended. Please install nnabla-ext-cuda101, nnabla-ext-cuda100, nnabla-ext-cuda90 or nnabla-ext-cuda80 instead.
The following nnabla CUDA extension packages have been deprecated and the PyPi repository has been closed.
- nnabla-ext-cuda91
- nnabla-ext-cuda92
The following "nnabla-ext-cuda" docker images have been deprecated.
- py37-cuda92
- py36-cuda92
- py27-cuda92
- py37-cuda92-v1.0.xx
- py36-cuda92-v1.0.xx
- py27-cuda92-v1.0.xx

We've decided to change nnabla's versioning policy to semantic versioning.
This change has been applied from version 1.1.0.

release-note-bugfix
+ [Set to use openib by default](https://github.com/sony/nnabla-ext-cuda/pull/419)
+ [Update nvidia/cuda base image](https://github.com/sony/nnabla-ext-cuda/pull/412)
+ [disable dl_nccl_generate when nccl_disabled](https://github.com/sony/nnabla-ext-cuda/pull/407)

release-note-build
+ [remove flatc build from Dockerfile](https://github.com/sony/nnabla-ext-cuda/pull/418)
+ [include cuda/cudnn libraries in wheel](https://github.com/sony/nnabla-ext-cuda/pull/415)
+ [Fix CUDA_VERSION_MINOR to various condition](https://github.com/sony/nnabla-ext-cuda/pull/414)
+ [add empty targz and merge cuda wheel](https://github.com/sony/nnabla-ext-cuda/pull/411)
+ [Prepare to build CUDA11.4 support](https://github.com/sony/nnabla-ext-cuda/pull/410)

release-note-core
+ [Optimize TransposeCuda with cuTENSOR](https://github.com/sony/nnabla-ext-cuda/pull/416)

release-note-utility
+ [Better error message in CUDNN.](https://github.com/sony/nnabla-ext-cuda/pull/409)


Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda110 For CUDA version 11.0.3 users


You can also install the cuda extension with specific versions from one of the following. [See also FAQ](https://nnabla.readthedocs.io/en/latest/python/install_on_linux.html#faq)

release-note-bugfix
+ [support libnccl.so.2](https://github.com/sony/nnabla-ext-cuda/pull/406)
+ [improve libmpi.so path searching method](https://github.com/sony/nnabla-ext-cuda/pull/399)

release-note-build
+ [Update protobuf to 3.19.4](https://github.com/sony/nnabla-ext-cuda/pull/405)
+ [use numpy 1.20.0 and later](https://github.com/sony/nnabla-ext-cuda/pull/404)
+ [Update gpg keys for centos based nvidia docker](https://github.com/sony/nnabla-ext-cuda/pull/402)
+ [update nvidia gpg key](https://github.com/sony/nnabla-ext-cuda/pull/401)
+ [Build flatc from latest upstream release](https://github.com/sony/nnabla-ext-cuda/pull/400)
+ [enable dynamic load nccl library](https://github.com/sony/nnabla-ext-cuda/pull/398)

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 For CUDA version 10.2 users


You can also install the cuda extension with specific versions from one of the following. [See also FAQ](https://nnabla.readthedocs.io/en/latest/python/install_on_linux.html#faq)

release-note-bugfix
+ [Fix: Slow Softmax backward when H and W are 1](https://github.com/sony/nnabla-ext-cuda/pull/395)

release-note-build
+ [Fix flatc build issue](https://github.com/sony/nnabla-ext-cuda/pull/397)

Install the latest nnabla by:

pip install nnabla
pip install nnabla-ext-cuda102 For CUDA version 10.2 users


You can also install the cuda extension with specific versions from one of the following. [See also FAQ](https://nnabla.readthedocs.io/en/latest/python/install_on_linux.html#faq)