Commits

Branch selector

 master

User selector

cheshire

Datepicker

All time

Commit History

Commits on Oct 18, 2019

• Automated g4 rollback of changelist 275470969.

  Show description for 96d7a52
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  96d7a52

  Copy full SHA for 96d7a52

• Crash on @tf.function(experimental_compile=True) in non-eager mode

  Show description for 2133490
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  2133490

  Copy full SHA for 2133490

Commits on Oct 17, 2019

• [XLA] [NFC] Move RendezvousKey and ParticipantData collective_ops_utils

  Show description for 4cd92b5
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  4cd92b5

  Copy full SHA for 4cd92b5

• [XLA GPU] [NFC] Pass AllReduce type and computation kind explicitly

  Show description for ee40c58
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  ee40c58

  Copy full SHA for ee40c58

• [XLA] [NFC] Move GetParticipatingReplicas to a helper library collective_ops_utils

  Show description for d21c713
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  d21c713

  Copy full SHA for d21c713

• [XLA Docs] Clarify limitations of xla.compile and experimental_compile=True

  Show description for 9dda088
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  9dda088

  Copy full SHA for 9dda088

Commits on Oct 16, 2019

• [XLA] [Docs] Document tf.function(experimental_compile=True)

  Show description for b82be1b
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  b82be1b

  Copy full SHA for b82be1b

Commits on Oct 15, 2019

• Log XLA compilation time with XLA_SCOPED_LOGGING_TIMER

  Show description for 42d095a
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  42d095a

  Copy full SHA for 42d095a

• [XLA] [Docs] Clarify replica_groups semantics

  Show description for 7af994f
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  7af994f

  Copy full SHA for 7af994f

• [SE] Merge `stream` and `stream_executor` targets

  Show description for 711f1b0
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  711f1b0

  Copy full SHA for 711f1b0

• [XLA] Drop memory limit from the redzone allocator used for gemm algorithm picking

  Show description for 3b71976
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  3b71976

  Copy full SHA for 3b71976

• [XLA] Rename `CrossReplicaSum` in documentation to `AllReduce`, specify `CrossReplicaSum` as a particular case

  Show description for dd37e79
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  dd37e79

  Copy full SHA for dd37e79

• [XLA] Buffer assignment: log total memory usage

  Show description for ba63ea3
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  ba63ea3

  Copy full SHA for ba63ea3

• Rename CrossReplicaSum XLA operation (which takes a custom computation) to AllReduce.

  Show description for ecade9b
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  ecade9b

  Copy full SHA for ecade9b

• [XLA Docs] Mention other frontends for XLA

  Show description for aec60bd
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  aec60bd

  Copy full SHA for aec60bd

Commits on Oct 11, 2019

• [XLA/GPU] Support more datatypes and more reduction operations for AllReduce

  Show description for 95f76d2
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  95f76d2

  Copy full SHA for 95f76d2

Commits on Oct 10, 2019

• Implement GetStream for all DeviceMemoryAllocator subclasses

  Show description for f616f06
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  f616f06

  Copy full SHA for f616f06

• [TF/XLA Bridge] [NFC] Extract blocks of code dealing with allocate_xla_tensors_ mode into separate functions

  Show description for c57c072
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  c57c072

  Copy full SHA for c57c072

• [XLA/GPU] Re-enable collective_ops_test

  Show description for 8757bc8
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  8757bc8

  Copy full SHA for 8757bc8

Commits on Oct 8, 2019

• [XLA/ROCm] Fix buildifier warning

  Show description for 77526f1
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  77526f1

  Copy full SHA for 77526f1

• [SE] [NFC] Change AsynchronousMemset to return Status

  Show description for b3aa334
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  b3aa334

  Copy full SHA for b3aa334

Commits on Oct 5, 2019

• [SE] [NFC] Introduce a macro for consistent error reporting from ROCm

  Show description for 41cc0ed
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  41cc0ed

  Copy full SHA for 41cc0ed

Commits on Oct 4, 2019

• [SE] [NFC] Change SynchronousMemset to return Status

  Show description for a3e7228
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  a3e7228

  Copy full SHA for a3e7228

• [SE] [NFC] Change LoadModuleFromHsaco to return port::Status

  Show description for 0abcb85
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  0abcb85

  Copy full SHA for 0abcb85

• [SE] [NFC] Change FuncSetCacheConfig to return Status

  Show description for c02773c
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  c02773c

  Copy full SHA for c02773c

• [SE] [NFC] Change FuncGetAttribute to return Status

  Show description for 48781b7
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  48781b7

  Copy full SHA for 48781b7

Commits on Oct 3, 2019

• [SE] Change GetDeviceName to return Status and not bool

  Show description for ea2c039
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  ea2c039

  Copy full SHA for ea2c039

• [SE] [NFC] Consistent CUDA error handling with a macro

  Show description for 9108d75
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  9108d75

  Copy full SHA for 9108d75

Commits on Oct 1, 2019

• [TF/XLA Bridge] [NFC] Reduce the amount of boilerplate required to create GraphOptimizationPassOptions

  Show description for 19156f6
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  19156f6

  Copy full SHA for 19156f6

• [TF/XLA Bridge] [NFC] Reduce the amount of boilerplate for getting-or-creating output tensor

  Show description for 86902a8
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  86902a8

  Copy full SHA for 86902a8

Commits on Sep 27, 2019

• [TF/XLA Bridge] Do not alias pass-through variables when run on XLA devices

  Show description for 1505381
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  1505381

  Copy full SHA for 1505381

Commits on Sep 26, 2019

• [TF/XLA Bridge] Alias pass-through parameters when the input graph contains no TF reference variables

  Show description for 6a20128
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  6a20128

  Copy full SHA for 6a20128

• [TF/XLA] Fix logic for aliasing output tensors to input

  Show description for 10c1e74
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  10c1e74

  Copy full SHA for 10c1e74

Commits on Sep 25, 2019

• Support dumping TF graphs to TEST_UNDECLARED_OUTPUTS_DIRS

  Show description for 9a544f5
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  9a544f5

  Copy full SHA for 9a544f5

Commits on Sep 20, 2019

• [XLA GPU] [NFC] Tiling codegen: minor simplifications and renames, add more comments.

  Show description for 18806ef
  

  cheshire

  authored and

  tensorflower-gardener

  committed
  18806ef

  Copy full SHA for 18806ef

Pagination

Previous

Next