122750a879f508342b53ee2eaecfde1656981280 |
|
31-Jul-2017 |
Peter Hawkins <phawkins@google.com> |
[SE] Make ExecutorCache thread-safe, change ExecutorCache::Insert to ExecutorCache::GetOrCreate. Add support for creating Executors for different device ordinals in parallel. [XLA] Create Executors in parallel. PiperOrigin-RevId: 163734988
/external/tensorflow/tensorflow/stream_executor/host/host_platform.cc
|
c31294351628addffc1cd89687fad0e81bac5a55 |
|
20-Jan-2017 |
A. Unique TensorFlower <gardener@tensorflow.org> |
Better StreamExecutor constructor Using std::unique for the implementation argument makes it clear that StreamExecutor is taking ownership of that argument. Change: 145107302
/external/tensorflow/tensorflow/stream_executor/host/host_platform.cc
|
a7b60a8206554270c1d066fd66242b1d90574a14 |
|
19-Jan-2017 |
A. Unique TensorFlower <gardener@tensorflow.org> |
Add hooks for StreamExecutor module initialization ordering Change: 145010633
/external/tensorflow/tensorflow/stream_executor/host/host_platform.cc
|
e4a4e922ed4be890f4273a7a276768dfefe80a4c |
|
05-Dec-2016 |
Peter Hawkins <phawkins@google.com> |
Add a Host platform to the stream executor library that runs code on the CPU. Avoid using a linker-initialized mutex in multi_platform_manager to work around a initialization/destruction ordering problem on Mac. Include the CUDA platform ID module even in non-CUDA builds. Change: 141045549
/external/tensorflow/tensorflow/stream_executor/host/host_platform.cc
|