Commits

Breadcrumbs

History for

onmrry-patch-1

User selector

All users

All time

Commit History

Commits on Feb 15, 2017

• Fix GetDsoHandle segfault (#7528)

  

  wrmsr

  authored and

  Vijay Vasudevan

  committed
  323e232

  Copy full SHA for 323e232

• FreeBSD compatibility (#7073)

  Show description for e75c0fc
  

  blodan

  authored and

  Vijay Vasudevan

  committed
  e75c0fc

  Copy full SHA for e75c0fc

Commits on Feb 14, 2017

• Merge commit for internal changes

  yifeif

  committed
  9067ac3

  Copy full SHA for 9067ac3

• [XLA:GPU] Cache kernels in `KernelThunk::ExecuteOnStream`

  Show description for 5799620
  tensorflower-gardener

  committed
  5799620

  Copy full SHA for 5799620

• Merge commit for internal changes

  

  Vijay Vasudevan

  committed
  9cce5df

  Copy full SHA for 9cce5df

• Fix code that ignores tensorflow::Status.

  Show description for bc225bf
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  bc225bf

  Copy full SHA for bc225bf

Commits on Feb 10, 2017

• Fix typo CUDNN_RETURN_IF_FAIL.

  theflofly

  committed
  7fc9bb7

  Copy full SHA for 7fc9bb7

• [StreamExecutor] Only check occupancy info at -v 2 and above.

  Show description for 30ef1b0
  

  Justin Lebar

  authored and

  tensorflower-gardener

  committed
  30ef1b0

  Copy full SHA for 30ef1b0

• Remove problematic SE_RETURN_STATUS_AS_BOOL macro

  Show description for 52fd3b9
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  52fd3b9

  Copy full SHA for 52fd3b9

Commits on Feb 9, 2017

• Fix double evaluation of macro argument that was causing duplicate CUDA batched GEMM calls.

  Show description for f439016
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  f439016

  Copy full SHA for f439016

Commits on Feb 7, 2017

• Adding support for non-fused Winograd algorithm from Cudnn 5.1.

  Show description for 764507e
  

  zheng-xq

  authored and

  tensorflower-gardener

  committed
  764507e

  Copy full SHA for 764507e

Commits on Feb 2, 2017

• Make dependencies on CUDA SDK libraries only if CUDA is configured

  Show description for edc1bf1
  tensorflower-gardener

  committed
  edc1bf1

  Copy full SHA for edc1bf1

• Merge changes from github.

  Show description for d1ba01f
  tensorflower-gardener

  committed
  d1ba01f

  Copy full SHA for d1ba01f

Commits on Jan 31, 2017

• Use map for platform-specific SE device options

  Show description for 6116e4b
  tensorflower-gardener

  committed
  6116e4b

  Copy full SHA for 6116e4b

• Remove Eigen/Core includes from public SE headers

  Show description for b4fa4ad
  tensorflower-gardener

  committed
  b4fa4ad

  Copy full SHA for b4fa4ad

Commits on Jan 27, 2017

• Default impl for StreamExecutorInterface::Memset

  Show description for 5480f32
  tensorflower-gardener

  committed
  5480f32

  Copy full SHA for 5480f32

• Remove "using StringPiece" from StreamExecutor lib/path.cc

  Show description for 41cb1b2
  tensorflower-gardener

  committed
  41cb1b2

  Copy full SHA for 41cb1b2

• Change the way the host platform stream blocks so that it is usable with a condition variable implementation that doesn't support waiting with a predicate.

  Show description for 8845387
  tensorflower-gardener

  committed
  8845387

  Copy full SHA for 8845387

Commits on Jan 25, 2017

• Stop using DSO loader for CUDA SDK libraries

  Show description for 191658d
  tensorflower-gardener

  committed
  191658d

  Copy full SHA for 191658d

Commits on Jan 24, 2017

• Add options argument for DNN activation

  Show description for eb65eff
  tensorflower-gardener

  committed
  eb65eff

  Copy full SHA for eb65eff

• Merge changes from github.

  Show description for 56fc883
  

  caisq

  authored and

  tensorflower-gardener

  committed
  56fc883

  Copy full SHA for 56fc883

Commits on Jan 20, 2017

• Add default ScopedDeviceMemory ctor in StreamExecutor

  Show description for 3ecafa3
  tensorflower-gardener

  committed
  3ecafa3

  Copy full SHA for 3ecafa3

• Better StreamExecutor constructor

  Show description for c312943
  tensorflower-gardener

  committed
  c312943

  Copy full SHA for c312943

Commits on Jan 19, 2017

• Add hooks for StreamExecutor module initialization ordering

  Show description for a7b60a8
  tensorflower-gardener

  committed
  a7b60a8

  Copy full SHA for a7b60a8

• Add convolve quantized ops to StreamExecutor API

  Show description for 598583c
  tensorflower-gardener

  committed
  598583c

  Copy full SHA for 598583c

• Addition of platform-specific key-value options to StreamExecutor DeviceOptions.

  Show description for 82542b6
  tensorflower-gardener

  committed
  82542b6

  Copy full SHA for 82542b6

• StreamExecutor param to specify cuDNN or TF padding

  Show description for b452d1c
  tensorflower-gardener

  committed
  b452d1c

  Copy full SHA for b452d1c

Commits on Jan 17, 2017

• Add several operations to the StreamExecutor API

  Show description for b398f90
  tensorflower-gardener

  committed
  b398f90

  Copy full SHA for b398f90

• Plumb port::Status through the internal synchronous memcopy routines.

  Show description for 21c0fa9
  tensorflower-gardener

  committed
  21c0fa9

  Copy full SHA for 21c0fa9

Commits on Dec 20, 2016

• Fix initialization order problem for mutexes on Mac GPU opensource build.

  Show description for 1466d8a
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  1466d8a

  Copy full SHA for 1466d8a

Commits on Dec 15, 2016

• Merge changes from github.

  Show description for 2e4869a
  

  martinwicke

  authored and

  tensorflower-gardener

  committed
  2e4869a

  Copy full SHA for 2e4869a

Commits on Dec 14, 2016

• Find libdevice files in .runfiles when TEST_SRCDIR is not set.

  Show description for 667a023
  

  Jingyue Wu

  authored and

  tensorflower-gardener

  committed
  667a023

  Copy full SHA for 667a023

Commits on Dec 9, 2016

• Merge changes from github.

  Show description for 1cb9689
  

  andrewharp

  authored and

  tensorflower-gardener

  committed
  1cb9689

  Copy full SHA for 1cb9689

Commits on Dec 8, 2016

• Remove undefined function GetRunfilesDirectory.

  Show description for 13d59e0
  

  Jingyue Wu

  authored and

  tensorflower-gardener

  committed
  13d59e0

  Copy full SHA for 13d59e0

• Separate the stream executor CUDA plugin into its own build target. This allows non --config=cuda builds to link the CUDA plugins if desired.

  Show description for 3e11249
  

  hawkinsp

  authored and

  tensorflower-gardener

  committed
  3e11249

  Copy full SHA for 3e11249