Commit graph

65 commits

Author SHA1 Message Date
Robert Scott 98fa0f29bc python3Packages.tensorflow: 2.4.1 -> 2.4.2 2021-07-22 23:55:38 +01:00
Jonathan Ringer 961cae37ed Revert "python3Packages.tensorflow: 2.4.0 -> 2.4.1"
This reverts commit 090a463e4d.
2021-06-16 18:39:39 -07:00
Jonathan Ringer e1b17b2d49 Revert "Updating relax-dependencies.patch"
This reverts commit 43362bd872.
2021-06-16 18:39:39 -07:00
Roger Qiu 43362bd872 Updating relax-dependencies.patch 2021-06-16 18:24:30 -07:00
Rakesh Gupta 090a463e4d python3Packages.tensorflow: 2.4.0 -> 2.4.1
python3Packages.tensorflow: 2.4.0 -> 2.4.1
2021-06-16 18:24:30 -07:00
Luke Granger-Brown f18734f8b8 python3Packages.tensorflow: update hashes for fetchConfigured 2021-06-07 20:59:01 +00:00
Daniël de Kok ef3322507a python3Packages.tensorflow: 2.4.0 -> 2.4.1
Changelog:
https://github.com/tensorflow/tensorflow/releases/tag/v2.4.1
2021-05-11 09:49:07 +02:00
Luke Granger-Brown d385b8f597 python3Packages.tensorflow: mark as requiring big-parallel
The Bazel build part of tensorflow builds can really use as many cores
as is available - to avoid blocking a small machine for literally hours
while it churns away, mark the build as big-parallel so we can schedule
it to run quickly on a machine with lots of cores.
2021-04-24 01:12:19 +00:00
Claudio Bley 6d8c431eb0 bazel: 3.7.1 -> 3.7.2
* update sha256 for gvisor, tensorflow{,WithCuda}, bazel-watcher, bazel-remote
  and envoy
2021-02-03 10:44:07 +01:00
Claudio Bley 71cda4f111 bazel: 3.3.1 -> 3.7.1
* remove python-shebang patch since it was integrated upstream[1]

[1]: https://github.com/bazelbuild/bazel/pull/11535/files
2021-01-29 21:53:25 +01:00
Markus Kowalewski 6dba41fbcb
mpi: use mpi attribute consistently as the default MPI implementations
Use the attribute mpi to provide a system wide default MPI
implementation. The default is openmpi (as before).
This now allows for overriding the MPI implentation by using
the overlay mechanism. Build all packages with mpich instead
of the default openmpi can now be achived like this:
self: super:
 {
   mpi = super.mpich;
 }

All derivations that have been using "mpi ? null" to provide optional
building with MPI have been change in the following way to allow for
optional builds with MPI:
{ ...
, mpi
, useMpi ? false
}
2021-01-23 12:15:13 +01:00
Jörg Thalheim 1ac4c06c08
Merge pull request #108923 from ndl/master 2021-01-11 17:47:28 +00:00
Profpatsch 4a7f99d55d treewide: with stdenv.lib; in meta -> with lib;
Part of: https://github.com/NixOS/nixpkgs/issues/108938

meta = with stdenv.lib;

is a widely used pattern. We want to slowly remove
the `stdenv.lib` indirection and encourage people
to use `lib` directly. Thus let’s start with the meta
field.

This used a rewriting script to mostly automatically
replace all occurances of this pattern, and add the
`lib` argument to the package header if it doesn’t
exist yet.

The script in its current form is available at
https://cs.tvl.fyi/depot@2f807d7f141068d2d60676a89213eaa5353ca6e0/-/blob/users/Profpatsch/nixpkgs-rewriter/default.nix
2021-01-11 10:38:22 +01:00
Alexander Tsvyashchenko 69d1e62b36 python3Packages.tensorflow_2: 2.3.2 -> 2.4.0
Also the following related changes:
* Removed Python 2 support because it's not supported by TF and related packages for a long time.
* Upgraded tensorboard and estimator packages to the required versions.
* Added extra plugins for tensorboard to support profiling.
* In the previous derivation versions, TF_SYSTEM_LIBS didn't have any effect because it was reset at repo fetching stage, so TF always used its own dependencies. Made TF_SYSTEM_LIBS actually work and fixed the errors caused by enabling it.
* Enabled tensorboard by default (but still keeping an option to disable it if needed).
2021-01-11 00:30:43 +01:00
Jörg Thalheim 370ffaaca7
tensorflow: 2.3.1 -> 2.3.2 2021-01-07 18:52:44 +01:00
Jörg Thalheim 1976d79627
tensorflow: drop 1.x
it no longer works without our default python version and needs
many patches to compile.
2021-01-07 18:52:44 +01:00
Timo Kaufmann a730888ce0 python.pkgs.tensorflow{,-bin,-estimator,-tensorboard}-2: Init at 2.1.0
Major breaking change from 1.x, so treating keeping both versions for now.

(adapted from 33f11be707)
(adapted from 9e8dea7986)
2020-03-30 13:58:28 +00:00
John Ericson d4de62dedd bazel_0_29: init at 0.29.1
Needed for Tensorflow 2.1
2020-03-30 04:30:40 +00:00
Andreas Rammhold 6c7b2ac58e pythonPackages.tensorflow-estimator_1_15_1: init at 1.15.1
TF 1.15 still needs an older version of the tensorflow-estimator
package.

(cherry picked from commit c539f937c5)
2020-03-28 03:05:54 +00:00
Jonathan Ringer ca6ccd4f4c pythonPackages.tensorflow: disable for python2.7 and 3.8
(cherry picked from commit b7bdf48e3f)
2020-03-25 16:43:28 +00:00
Dylan Simon 436a4bace6 tensorflow: 1.15.1 -> 1.15.2
(cherry picked from commit 0a5ec494b574ccd3db3b234c7d6637d127bd16da)
(cherry picked from commit 1e769bdd2a)
2020-03-25 16:43:10 +00:00
Andreas Rammhold 08f2ed7b02 pythonPackages.tensorflow: use bazel_0
This allows us to get rid of the compatibility hacks that we had to add
(tf-1.15-bazel-1.0.patch) and also fixes #77626.

(cherry picked from commit c7adb4ee72)
2020-03-25 16:43:02 +00:00
Andreas Rammhold 4e7dac0a56 tensorflow: update fixed output hashes
(cherry picked from commit 642bea6cec)
2020-03-25 16:42:53 +00:00
Michael Lingelbach a0b992fdf8 tensorflow: 1.15.0 -> 1.15.1
* Apply glibc 2.3 patch
* build tensorflow with bazel_1
* Bump openssl version to 1.1

(cherry picked from commit 34296980d1)
2020-03-25 16:42:27 +00:00
Pavol Rusnak 9a982d3540
python.pkgs.tensorflow: fix build with cudatoolkit 10.2 2020-01-06 13:46:00 +01:00
Robin Gloster 313da176d3
treewide: NIX_*_FLAGS -> string 2019-12-31 00:16:46 +01:00
Dmitry Kalinkin 9c42df8d3e
python3Packages.tensorflow-build: enable on darwin 2019-11-27 08:34:25 -05:00
Timo Kaufmann 8e382a7ca7 python.pkgs.tensorflow: mark as broken for python2 2019-11-13 16:18:04 +01:00
Timo Kaufmann e4378d4641 python.pkgs.tensorflow: add more sanity checks 2019-11-13 16:18:01 +01:00
Timo Kaufmann 6832f1da6a python.pkgs.tensorflow: add functools32 dependency
Now needed at runtime with the python2 build.
2019-11-13 15:25:19 +01:00
Timo Kaufmann 4fb7831cb4 python.pkgs.tensorflow: fix build with newer gast 2019-11-12 23:43:10 +01:00
Timo Kaufmann 1fbd307880 python.pkgs.tensorflow: fix build with bazel 1.0
The tensorflow build was broken by the bazel 1.0 update in
73eb01bde000c34b54764f6f2b9fd1d89a54ccfa.
2019-11-12 23:43:07 +01:00
Timo Kaufmann 152d2fd22c python.pkgs.{tensorflow,tensorflow-estimator,tensorflow-tensorboard}: 1.14.0 -> 1.15.0 2019-11-12 23:42:47 +01:00
Nikolay Amiantov 1c429acbff tensorflow: add OpenGL path to find libcudart 2019-10-02 10:33:31 +03:00
Nikolay Amiantov 4947ddf347 libtensorflow: add meta 2019-10-02 10:33:31 +03:00
Nikolay Amiantov 46b7933d9a tensorflow: fix CUDA build using wrong GCC 2019-10-02 10:33:31 +03:00
Vladimír Čunát f21211ebfe
Merge branch 'master' into staging 2019-09-02 23:25:24 +02:00
Nikolay Amiantov e0795aa5e8 tensorflow: fix python2 build
Avoid using PYTHONPATH, switch to python envs instead.
2019-08-31 21:02:55 +03:00
volth 08f68313a4 treewide: remove redundant rec 2019-08-28 11:07:32 +00:00
Nikolay Amiantov 71095460b9 tensorflow: fix build
You know, Tensorflow.
2019-08-27 02:43:48 +03:00
Nikolay Amiantov 3b3618ea3c buildBazelPackage: remove top-level symlinks
They always can be regenerated during the actual build, and they are sometimes
random, e.g in Tensorflow;

platforms -> NIX_BUILD_TOP/tmp/install/35282f5123611afa742331368e9ae529/_embedded_binaries/platforms
2019-08-19 11:23:17 +03:00
Nikolay Amiantov 19cdfe87f9 python.pkgs.tensorflow: update dependencies hash 2019-07-31 13:28:44 +03:00
Timo Kaufmann d30ec1a0ef python.pkgs.tensorflow: fix optimization flags
Now need to be passed in the configure phase.

abbradar: Don't change CUDA build hash.
2019-07-31 13:28:44 +03:00
Nikolay Amiantov 019c13616b tensorflow: cleanup and merge with libtensorflow
This merges work done by yorickvP and timokau in #63208 and #63616 respectively.
Now the derivation builds both libtensorflow and the Python package and puts them into
different outputs.

Quite a bit of improvements were done on the top, including:

* Use official tag revision as source, not a branch;
* Use all system libraries possible (before only one was actually used);
* Move various environment variables to the derivation itself from hooks;
* Use source Python build instead of wheel build to ensure fixup hooks do their important jobs on libraries;
* And more that I forgot!
2019-07-31 13:28:44 +03:00
Timo Kaufmann 3df4e2df95 tensorflow: re-enable build from source [WIP] 2019-07-31 13:28:15 +03:00
volth 5ad79dc4bb compilation for particular x86_64 architecture 2019-04-30 14:28:04 +00:00
Frederik Rietdijk ced21f5e1a pythonPackages: remove name attribute`
The `buildPython*` function computes name from `pname` and `version`.
This change removes `name` attribute from all expressions in
`pkgs/development/python-modules`.

While at it, some other minor changes were made as well, such as
replacing `fetchurl` calls with `fetchPypi`.
2018-06-23 18:14:26 +02:00
Nikolay Amiantov 650d8c93a1 tensorflow: add a binary wheel release for Darwin 2018-04-05 15:00:18 +03:00
Nikolay Amiantov 94ebc13a6a tensorflow: disable XLA without CUDA
This configuration is sadly broken.
2018-02-27 03:16:35 +03:00
Nikolay Amiantov 8aeea493a5 python.pkgs.tensorflow: 1.3 -> 1.5 2018-02-21 00:50:24 +03:00