This fixes a potential race condition and segfault, where the response
buffer could be deallocated with an RPC in flight. This would happen
if a step was cancelled before its master blocked on the RunGraph RPCs.
Change: 124255996

* Updated iOS settings (#2694)

* Added linker options to armv7 in iOS makefile

* Updated Eigen version in Xcode iOS examples

* Updated iOS link flags

* Fixes for Raspberry Pi compilation (#2701)

* Added linker options to armv7 in iOS makefile

* Updated Eigen version in Xcode iOS examples

* Updated iOS link flags

* Added support for Raspberry Pi

* Updated Pi documentation and Eigen version in makefile (#2743)

* Updated Pi compilation documentation

* Updated Eigen version in makefile

* Updated documentation link

* Updated Pi documentation

* Updated Pi build settings

* Fix for build issue #2742; (#2749)

(1) Additionally added linking against pthread for linux machines

* Missing space; (#2762)

(1) bash file check will fail without the space;

* Updated Eigen include paths for iOS examples (#2841)

* Updated Eigen hashes

* Updated Eigen paths to a permanent link

* Merged Eigen header changes

* Merged Eigen header changes

Shrink iOS binaries

types can be avoided on mobile platform.
Change: 124172890

macros, and change the call-many macros to use those.

This will let us change some other kernels to use TF_CALL*, for cases where
no existing subset of types is what the kernel wants. This can help size on
IOS.
Change: 124166009

Cherry pick dist/tf-tensorboard.html update onto r0.9

Change: 124375656

parameterized_docker_build includes three CUDA capabilities

Avoid too much duplication (still, three sections are almost copied) when specifying installation instructions. Added 3.5 linux binaries.

Replace assertEquals in graph_io_test; disable a session_bundle test

assertEquals is deprecated and causes Python 3.5 test failures.

temporarily disable contrib/session_bundle

The genrule makes problems in integration testing. We disable it
for now and fix it after 0.9 release.

added 0.9 release notes.
added contributors
Removed completed items, added new features on the horizon

Bump version to 0.9.0rc0

Branch 124012080

* Added Raspberry Pi cross-compilation support to makefile

* Fixed makefile linking problem with Pi

* Added documentation for Pi compilation

* Added live camera iOS example

* Updated Eigen version in makefile, and added more iOS compile flags

* Added licenses to makefile sh scripts, fixed simple iOS example setting

Change: 124012080

- Add a ReadUpTo method to ReaderInterface
- Add ReadUpToLocked to ReaderBase w/ default implementation.
  Readers may override this method to achieve better batched
  I/O performance.
Change: 124012016

Change: 124007278

This takes previously generated code, and includes it in the
repository. The main advantage of doing this is that we can specialize
the deserialization routines for various protobuf types that tend to
be large, and thereby avoid the problem where we brush up against the
default protobuf limits.

Fixes #2233.
Change: 124007049

Cleanup: Get rid of unused copy of simple threadpool implementation. The implementation is identical to the simple threadpool in Eigen (can be selected by not defining EIGEN_USE_NONBLOCKING_THREAD_POOL), which also has had all the same tracing calls added via the hooks in EigenEnvironment.
Change: 124006070

Change: 124005557

Change: 124003810

Change: 124003630

* conv_grad_ops.cc
* cwise_op_inverse.cc
* cwise_op_squared_difference.cc
* resize_bilinear_op.cc
Change: 123999923

Add x permission back to test_tutorials.sh

Change: 123998188

* Added Raspberry Pi cross-compilation support to makefile

* Fixed makefile linking problem with Pi

* Added documentation for Pi compilation

* Added live camera iOS example

* Updated Eigen version in makefile, and added more iOS compile flags

Changing API for monitors step_begin to only take step and return list of tensors to run. Support distributed training in EveryN monitor.
Change: 123996408

* Enable GPU for L2Loss float64

* Enable GPU for L2Loss float64