Commit Graph

219 Commits

Author SHA1 Message Date
Mihai Maruseac
5fbe995c00 Prepare build jobs for 1.13.2 patch release
PiperOrigin-RevId: 255718561
2019-07-01 10:34:50 -07:00
A. Unique TensorFlower
cfccbdb8c4 Set --host_force_python=PY2 when using python2
Fixes #29220

PiperOrigin-RevId: 254793086
2019-06-24 11:38:01 -07:00
Sanjoy Das
f3fab7c406 Link in XLA for open source OSX builds by default
PiperOrigin-RevId: 254235029
2019-06-20 11:26:26 -07:00
Mark Daoust
44000ad64a add _TF_{MIN,MAX}_BAZEL_VERSION constants, to make these easier to find.
PiperOrigin-RevId: 253803878
2019-06-18 09:31:38 -07:00
Anna R
aedf674e90 Add --config=v2 to build-from-source documentation and configure.py.
PiperOrigin-RevId: 252864895
2019-06-12 12:23:24 -07:00
A. Unique TensorFlower
c3260eb5ab Bump max Bazel version to 0.26.1
PiperOrigin-RevId: 252051374
2019-06-07 08:21:32 -07:00
Gaurav Jain
040546fdef Upgrade Bazel version to 0.26.0
PiperOrigin-RevId: 250559259
2019-05-29 14:20:45 -07:00
A. Unique TensorFlower
c33055c442 Upgrade Bazel version to 0.25.3
PiperOrigin-RevId: 249794093
2019-05-24 01:39:59 -07:00
A. Unique TensorFlower
fef5167133 Wrap some (CUDA, TensorRT, clang, SYCL) --action_envs into --configs: the .tf_configure.bazelrc only sets the --configs, and the .bazelrc maps that to --action_envs.
PiperOrigin-RevId: 249618180
2019-05-23 04:24:32 -07:00
A. Unique TensorFlower
8d7696a800 Removes min Bazel version check when building for Apple platforms.
PiperOrigin-RevId: 248742379
2019-05-17 10:23:42 -07:00
A. Unique TensorFlower
3ad37ad68a Remove unnecessary duplicate config for build and test in .tf_configure.bazelrc: test inherits from build, see https://docs.bazel.build/versions/master/guide.html#option-defaults
PiperOrigin-RevId: 248330560
2019-05-15 07:42:33 -07:00
A. Unique TensorFlower
9262f195e7 Bump max Bazel version to 0.25.2
Also removed a workaround

PiperOrigin-RevId: 248095796
2019-05-14 01:59:52 -07:00
A. Unique TensorFlower
28fc9cccd1 Fix ./configure trying to find TensorRT when it shouldn't.
PiperOrigin-RevId: 246197213
2019-05-01 15:59:23 -07:00
Justin Lebar
7b57c5a02c Attempt to fix Windows build by explicitly defining NOGDI.
My hypothesis is that WIN32_LEAN_AND_MEAN is not sufficient for whatever header
is causing StreamExecutor to pull in GDI.

PiperOrigin-RevId: 245998050
2019-04-30 13:31:39 -07:00
A. Unique TensorFlower
b2e7f671da Switch from TF_TENSORRT_VERSION to TF_NEED_TENSORRT environment variable to determine whether to build with TF-TRT.
Store fallback env variables for CUDA configs in bazelrc as well.

PiperOrigin-RevId: 245961133
2019-04-30 09:26:48 -07:00
Guangda Lai
ad5ef82ea3 Automated rollback of commit f5b4e0e0ea
PiperOrigin-RevId: 245496956
2019-04-26 17:14:04 -07:00
Guangda Lai
f5b4e0e0ea Add back action_env TENSORRT_INSTALL_PATH so TF-TRT is built by default in TF gpu builds.
PiperOrigin-RevId: 245462577
2019-04-26 13:27:16 -07:00
Mihai Maruseac
4db860f9f0 Set max bazel version to the latest existing 0.24.1
PiperOrigin-RevId: 244365126
2019-04-19 08:57:34 -07:00
R S Nikhil Krishna
05e348bf7b
Update docstrings in configure.py
HDFS is now supported through bazel build config and not through a flag in this file (`configure.py`).

However, the docstrings still refer to it. Replaced it with `TF_NEED_CUDA`, which is still a flag set in this script
2019-04-18 15:58:12 +05:30
A. Unique TensorFlower
407a4f3b9d Bump the minimal Bazel version to 0.24.1
We'll have to bump the minimal Bazel version to 0.24.1 because google_cloud_cpp v0.8.1 doesn't work with Bazel 0.23.1.

PiperOrigin-RevId: 243954431
2019-04-17 01:30:14 -07:00
A. Unique TensorFlower
0cb8a070d9 Suggest default CUDA/cuDNN/TensorRT versions after initial auto-detection.
Reduce spew on error message when some file was not found.

PiperOrigin-RevId: 243810113
2019-04-16 08:22:21 -07:00
A. Unique TensorFlower
651b0bcbbc Automatically detect CUDA config during ./configure on the first attempt. If that fails, ask for more detail.
PiperOrigin-RevId: 243794789
2019-04-16 06:19:25 -07:00
A. Unique TensorFlower
8e9ca6617b Change how ./configure and bazel finds CUDA paths.
Use a single python script (third_party/gpus/find_cuda_config.py) from configure.py and the different *_configure.bzl scripts to find the different CUDA library and header paths based on a set of environment variables.

PiperOrigin-RevId: 243669844
2019-04-15 12:55:59 -07:00
A. Unique TensorFlower
6a059c00dd Updates TensorFlowLiteC podspec.
PiperOrigin-RevId: 242194312
2019-04-05 15:02:02 -07:00
Jared Duke
a0104b7a5a Tweak NDK API level parsing for Android
Allow setting the level via $ANDROID_NDK_API_LEVEL, rather than
hardcoding to the current NDK version. Also increase the default
NDK API level to 18.

PiperOrigin-RevId: 241981236
2019-04-04 12:36:28 -07:00
A. Unique TensorFlower
2b9c299716 Bump max bazel version to 0.24.1
Bazel 0.24.1 is now released: https://github.com/bazelbuild/bazel/releases/tag/0.24.1

PiperOrigin-RevId: 241955827
2019-04-04 10:42:53 -07:00
Austin Anderson
87ea41d023 Fix configure.py to properly compare "X.Y.Z" with "X.Y"
Right now, "0.24" is treated as lower than "*.*.*" because of the odd comparison method that adds digits to each existing section. This change converts "0.24" to "0.24.0" to fix that.

This will probably need to be pulled into r2.0.

PiperOrigin-RevId: 241950768
2019-04-04 10:18:24 -07:00
A. Unique TensorFlower
5001ac0ae1 Internal change.
PiperOrigin-RevId: 241777236
2019-04-03 12:05:15 -07:00
A. Unique TensorFlower
c7278e4620 PR #26752: Bump abseil-py dep and min Bazel version
Imported from GitHub PR #26752

This makes builds not fail with --incompatible_remove_old_python_version_api enabled, which will become the default in Bazel 0.25. Fixes #26616.

Copybara import of the project:

  - a3b9e5cde69beb5b1ceebe9a4a20c352bf320290 Bump abseil-py dep and min Bazel version by Jon Brandvein <brandjon@google.com>
  - d189be43b2bf4bcd229d4fe23bc95870a196ff79 Merge a3b9e5cde69beb5b1ceebe9a4a20c352bf320290 into 9ab5b... by Jon Brandvein <brandjon@google.com>

PiperOrigin-RevId: 240409003
2019-03-26 13:05:20 -07:00
A. Unique TensorFlower
8a9cd67c69 Bump max bazel version to 0.24.0
Bazel 0.24.0 is now released: https://github.com/bazelbuild/bazel/releases/tag/0.24.0

Tested at https://buildkite.com/bazel/bazel-at-head-plus-downstream/builds/867#e6631eba-578c-4cd5-8275-0310317b7bc8

PiperOrigin-RevId: 240395484
2019-03-26 11:54:13 -07:00
A. Unique TensorFlower
e9797fdf8a Exclude setting BAZEL_USE_CPP_ONLY_TOOLCHAIN=1 for ios build
PiperOrigin-RevId: 239826351
2019-03-22 11:28:19 -07:00
A. Unique TensorFlower
76e879d1c1 Adds new podspecs for the TensorFlow Lite iOS libraries.
PiperOrigin-RevId: 239748653
2019-03-22 00:31:14 -07:00
A. Unique TensorFlower
7bd86377de Set BAZEL_USE_CPP_ONLY_TOOLCHAIN=1 for mac build
For working around https://github.com/bazelbuild/bazel/issues/7607#issuecomment-474380287

PiperOrigin-RevId: 239684343
2019-03-21 15:44:45 -07:00
A. Unique TensorFlower
ed297347fd Adds a print statement alerting developers that Bazel version 0.23 or later is required for building Apple and Swift targets on Apple platforms.
PiperOrigin-RevId: 238678544
2019-03-15 12:02:44 -07:00
Guangda Lai
bcd701a0d4 Fix a bug in the configure script that disables all GPU tests when GPU build is enabled.
PiperOrigin-RevId: 238155623
2019-03-12 21:14:39 -07:00
A. Unique TensorFlower
4a866ad090 Bump max bazel version for TF to 0.23.2.
https://github.com/bazelbuild/bazel/issues/7667

PiperOrigin-RevId: 238045795
2019-03-12 10:50:41 -07:00
A. Unique TensorFlower
8dc2d0eeda Change compute capability < 3.5 check from error into warning. See issue #26394.
PiperOrigin-RevId: 237971151
2019-03-12 01:46:54 -07:00
A. Unique TensorFlower
e7a30e0791 Bump max bazel version for TF to 0.23.1.
Passed with 0.23.1rc1 at https://buildkite.com/bazel/bazel-at-head-plus-downstream/builds/833#8d13d2b7-28cf-4576-a00a-d6dd6ea3edab

PiperOrigin-RevId: 236876533
2019-03-05 11:17:39 -08:00
A. Unique TensorFlower
e89c7daf6c Bump max bazel version for TF to 0.23.1.
Passed with 0.23.1rc1 at https://buildkite.com/bazel/bazel-at-head-plus-downstream/builds/833#8d13d2b7-28cf-4576-a00a-d6dd6ea3edab

PiperOrigin-RevId: 236362086
2019-03-01 13:03:42 -08:00
TensorFlower Gardener
345cccf7b8 Merge pull request #25447 from rongjiecomputer:fix_win_build
PiperOrigin-RevId: 236176304
2019-02-28 15:22:59 -08:00
A. Unique TensorFlower
a6bf9c8476 Add posix implementation for platform/numa.h functions, relying
on hwloc.

PiperOrigin-RevId: 235742876
2019-02-26 10:36:09 -08:00
Gunhan Gulsoy
fff6696b9f Let configure script set TF test options in bazelrc
PiperOrigin-RevId: 234881690
2019-02-20 15:34:00 -08:00
Austin Anderson
32202dc07c
Adjust wording in warning 2019-02-19 10:46:27 -08:00
P Sudeepam
765ceda2f0 Remove an accidental tab and quote 2019-02-19 14:46:38 +05:30
P Sudeepam
5209356cdb Remove the warning_flag and use the all_valid flag to provide the warning for invalid compute capabilities 2019-02-19 14:46:38 +05:30
P Sudeepam
6313548a1f modify the if condition that sets the warning flag for compute capabilities 2019-02-19 14:46:38 +05:30
P Sudeepam
d910c0289b Modify configure.py to provide a warning if CUDA compute capabilities are less than 3.5 2019-02-19 14:46:38 +05:30
A. Unique TensorFlower
93e707396e Updates the configure Python script to support building Bazel rules on Apple platforms.
PiperOrigin-RevId: 234053872
2019-02-14 17:00:09 -08:00
TensorFlower Gardener
abb763b5ae Merge pull request #25335 from lgeiger:deprecated-bazel-option
PiperOrigin-RevId: 232765856
2019-02-06 16:23:03 -08:00
Gunhan Gulsoy
91da898cb6 Bump max bazel version for TF.
I just verified that we can build using bazel 0.22

PiperOrigin-RevId: 232356025
2019-02-04 13:36:44 -08:00