Age | Commit message (Collapse) | Author |
|
libibverbs was merged into rdma-core.
References
----------
https://lists.archlinux.org/pipermail/aur-requests/2018-October/027305.html
|
|
|
|
opencl is now disabled due to an upstream build error [1].
References
----------
[1] https://github.com/pytorch/pytorch/issues/9990
|
|
This package now provides the non-cuda version (as known as the
'cpy only' build).
If you want caffe2 with cuda support, use package caffe2-cuda.
This new package naming schema will better reflect the package
contents. This conforms to tensorflow package naming from the
official repositories.
|
|
|
|
|
|
cuda 9.2 compiler is gcc7, but compilation fails with it. It works
with gcc6 but it's not on the official repositories.
gcc54 is on the official repositories, so let's use it.
|
|
gcc54 is now a dependency for cuda, sot it will be already
pulled. Letting gcc5 (as it was until now) will cause a
package conflict with gcc54.
References
----------
https://git.archlinux.org/svntogit/community.git/commit/trunk?h=packages/cuda&id=2b0be9b7390fa897edb6e4f44224eaf8619399c9
|
|
Caffe2 source code moved to PyTorch repository. It now resides
into pytorch releases.
Note
----
gcc5 is currently on the [community] repository. It was
moved from the AUR to [community] due to CUDA. Removed
gcc5 from the AUR block on makedepends.
|
|
Unfortunatelly, the current available stable version of caffe2
does not build with the current cuda 9.1 and gcc6 from the
official repositories. This prevented the package to be built
for a long time.
Details here:
https://github.com/caffe2/caffe2/issues/1459
https://github.com/caffe2/caffe2/issues/1636
But it is possible to build with gcc5 from the AUR.
This commit adds build fixes for cuda 9.x, including the use
of gcc5. Be warned that that gcc5 from the AUR takes a lot of
time to compile.
Important notes
---------------
1) The Caffe2 source code moved to PyTorch repository, but
currently there is no stable version containing Caffe2 there.
It is yet to be released.
2) The current available stable version of Caffe2 does not have
python3 support. Python3 support is available on the git
master branch. If you want python3 support right now, please
use caffe2-git or caffe2-cpu-git package.
|
|
|
|
This required the removal off gloo support.
Hopefully gloo can be enabled in a future caffe2 version, since
it is working on git master (caffe2-git).
Also removed the now unneeded nccl cmake lines.
|
|
The newly released NCCL version 2 requires registration at the
NVIDIA developer webiste and a manual download. As this breaks
automated builds by AUR helpers, NCCL was disabled by default in
this package.
|
|
|
|
Adding python-future fixes an error when importing the caffe2
python modules. Thanks to AUR user wangqr for pointing this.
|
|
This version adds support for the newly released cudnn 7.0.
|
|
Note about cudnn:
Caffe2 currently does not compile with the newly released cudnn 7.0.
It currently requires cudnn < 7.0 and builds fine with cudnn 6.0.21.
https://github.com/caffe2/caffe2/issues/1025
|
|
|
|
|
|
|
|
|
|
caffe2 uses sources from many third party components during the
building process. It's better to use the same third party versions
that were used by upstream caffe2 developers at time of a given
caffe2 stable release. This can avoid compile errors and reduce
compile warnings. These versions are given by third party git
commit hashes that can be found on the caffe2 git repository.
The exact expected third party version (git commits hashes) at the
time of a given caffe2 stable release can be obtained by:
- cloning caffe2 git repository
- 'git checkout <caffe2 stable release commit hash>'
- 'git submodule status' to show the needed commits
- use 'git diff' to get previous git commit hashes of submodules
marked with a '+' (modified hashes/versions).
Some third party git commit hashes coincides with a corresponding
stable release. In this case we do not need to use git sources
since the '.tar.gz' file of the third party stable release is the
same snapshot of the given git commit hash.
|
|
|
|
Compiling with opencv currently is giving linking errors
with libtbb.so (intel-tbb, enabled in opencv).
Temporarily disabled opencv until it can be fixed.
|
|
|
|
Version 0.1 is a pre-release from august/2015.
Since pthreadpool is a nnpack dependency and nnpack has only a git
master version (no releases yet), using pthreadpool git master can
avoid possible problems in nnpack.
|
|
cudnn has been moved from AUR to [community].
|
|
cudnn already pulls cuda. The current version of cuda (8.0) already
pulls gcc5. But strictly adding them the dependencies can avoid
some problems in the future.
|
|
|
|
|