Commits

Breadcrumbs

History for

onmaster

User selector

All users

All time

Commit history

Commits on Jan 10, 2018

• TensorFlow for NVIDIA Tegra devices with CUDA support (#14167)

  Show description for 589a2d4
  

  lihanchen

  authored and

  andrewharp

  committed
  589a2d4

  Copy full SHA for 589a2d4

Commits on Jan 9, 2018

• [XLA:GPU] Warn if ptxas or the driver JIT has known bugs.

  Show description for fffcacb
  

  Justin Lebar

  authored and

  tensorflower-gardener

  committed
  fffcacb

  Copy full SHA for fffcacb

Commits on Dec 29, 2017

• Merge changes from github.

  Show description for 20765b3
  

  Patrick Nguyen

  authored and

  tensorflower-gardener

  committed
  20765b3

  Copy full SHA for 20765b3

Commits on Dec 22, 2017

• Merge changes from github.

  Show description for e4532d2
  tensorflower-gardener

  committed
  e4532d2

  Copy full SHA for e4532d2

• Adds support in stream executor interface to update the scratch allocator used with a cuFFT plan. This enables plan reuse without requiring we keep the scratch allocation alive between executions.

  Show description for 4c0adf2
  tensorflower-gardener

  committed
  4c0adf2

  Copy full SHA for 4c0adf2

Commits on Dec 21, 2017

• Fix padding for int8 fused convolution.

  Show description for c7a05f4
  

  Jingyue Wu

  authored and

  tensorflower-gardener

  committed
  c7a05f4

  Copy full SHA for c7a05f4

Commits on Dec 18, 2017

• [StreamExecutor] Change "variance" to "inv_var" in BatchNormalizationBackward.

  Show description for 483e439
  

  Justin Lebar

  authored and

  tensorflower-gardener

  committed
  483e439

  Copy full SHA for 483e439

• [StreamExecutor] Allow null batch_mean/batch_var in calls to BatchNormalizationForward.

  Show description for 01da208
  

  Justin Lebar

  authored and

  tensorflower-gardener

  committed
  01da208

  Copy full SHA for 01da208

Commits on Dec 15, 2017

• Rename StreamExecutorInterface::BlockHostUntilDoneWithStatus to BlockHostUntilDone.

  Show description for 2df1a5c
  tensorflower-gardener

  committed
  2df1a5c

  Copy full SHA for 2df1a5c

Commits on Dec 13, 2017

• Rename Stream::BlockHostUntilDoneWithStatus to BlockHostUntilDone.

  Show description for 70062d1
  tensorflower-gardener

  committed
  70062d1

  Copy full SHA for 70062d1

Commits on Dec 12, 2017

• Use BlockHostUntilDoneWithStatus in various places.

  Show description for 383a322
  tensorflower-gardener

  committed
  383a322

  Copy full SHA for 383a322

Commits on Dec 11, 2017

• Fix mismatched argument comments to match parameter names

  Show description for 309f7e2
  tensorflower-gardener

  committed
  309f7e2

  Copy full SHA for 309f7e2

Commits on Dec 9, 2017

• Replace StreamExecutorInterface::BlockHostUntilDone with BlockHostUntilDoneWithStatus

  Show description for 3764127
  tensorflower-gardener

  committed
  3764127

  Copy full SHA for 3764127

Commits on Dec 6, 2017

• [StreamExecutor] When a kernel launch fails, print the kernel's name.

  Show description for 3d895eb
  

  Justin Lebar

  authored and

  tensorflower-gardener

  committed
  3d895eb

  Copy full SHA for 3d895eb

Commits on Dec 4, 2017

• [StreamExecutor] Add UnqueryableDeviceParams for all nvidia GPUs.

  Show description for 4a8e27e
  

  Justin Lebar

  authored and

  tensorflower-gardener

  committed
  4a8e27e

  Copy full SHA for 4a8e27e

Commits on Nov 27, 2017

• Delete trailing whitespace

  Show description for 191825e
  tensorflower-gardener

  committed
  191825e

  Copy full SHA for 191825e

Commits on Nov 23, 2017

• Minor cleanup: remove unnecessary GetCudaContext.

  Show description for f25abbf
  tensorflower-gardener

  committed
  f25abbf

  Copy full SHA for f25abbf

Commits on Nov 22, 2017

• Merge changes from github.

  Show description for b1d8c59
  

  yifeif

  authored and

  tensorflower-gardener

  committed
  b1d8c59

  Copy full SHA for b1d8c59

• Automated g4 rollback of changelist 176615107

  Show description for e70c009
  

  yifeif

  authored and

  tensorflower-gardener

  committed
  e70c009

  Copy full SHA for e70c009

• Merge changes from github.

  Show description for c6d603f
  

  yifeif

  authored and

  tensorflower-gardener

  committed
  c6d603f

  Copy full SHA for c6d603f

Commits on Nov 11, 2017

• Include memory size info in error message.

  Show description for 8d15389
  

  protoget

  authored and

  aselle

  committed
  8d15389

  Copy full SHA for 8d15389

• [StreamExecutor] LOG(ERROR) the driver version when cudnnCreate fails.

  Show description for f9c7db1
  

  Justin Lebar

  authored and

  aselle

  committed
  f9c7db1

  Copy full SHA for f9c7db1

Commits on Nov 7, 2017

• Clean up kernels cached by CUDAExecutor.

  Show description for c070ab2
  

  Artem-B

  authored and

  tensorflower-gardener

  committed
  c070ab2

  Copy full SHA for c070ab2

Commits on Nov 3, 2017

• Support Cudnn RNN Fp16

  Show description for 509d0f2
  

  protoget

  authored and

  tensorflower-gardener

  committed
  509d0f2

  Copy full SHA for 509d0f2

Commits on Nov 2, 2017

• Merge changes from github.

  Show description for 8891788
  

  andrewharp

  authored and

  tensorflower-gardener

  committed
  8891788

  Copy full SHA for 8891788

Commits on Oct 31, 2017

• Prefer cubin over PTX when we launch CUDA kernels.

  Show description for 8e732a3
  

  Artem-B

  authored and

  tensorflower-gardener

  committed
  8e732a3

  Copy full SHA for 8e732a3

Commits on Oct 18, 2017

• Use Windows compatible string comparisons for setting cuda device flags.

  Show description for 9b3233d
  tensorflower-gardener

  committed
  9b3233d

  Copy full SHA for 9b3233d

Commits on Oct 17, 2017

• Add environment variable to enable setting of CUDA context flags.

  Show description for 1091336
  tensorflower-gardener

  committed
  1091336

  Copy full SHA for 1091336

Commits on Oct 10, 2017

• Add an env-var to choose between FP16 and FP32 as the internal compute type for conv when input data is FP16.

  Show description for 70e2cbf
  

  Yangzihao Wang

  authored and

  tensorflower-gardener

  committed
  70e2cbf

  Copy full SHA for 70e2cbf

• Make error message more explicit when running FusedConv2DBiasActivationOp with type int8 on a GPU that doesn't support it.

  Show description for 2cdd064
  tensorflower-gardener

  committed
  2cdd064

  Copy full SHA for 2cdd064

Commits on Oct 6, 2017

• Merge changes from github.

  Show description for 5eaefba
  

  Frank Chen

  authored and

  tensorflower-gardener

  committed
  5eaefba

  Copy full SHA for 5eaefba

Commits on Oct 5, 2017

• Add dilation rates support for ConvolutionDescriptor...

  Show description for b0e751a
  

  Yangzihao Wang

  authored and

  tensorflower-gardener

  committed
  b0e751a

  Copy full SHA for b0e751a

Commits on Sep 27, 2017

• Add float16 support to tf.nn.fused_batch_norm on the GPU.

  Show description for 759690f
  

  reedwm

  authored and

  tensorflower-gardener

  committed
  759690f

  Copy full SHA for 759690f

Commits on Sep 26, 2017

• Merge changes from github.

  Show description for e2e3a94
  

  caisq

  authored and

  tensorflower-gardener

  committed
  e2e3a94

  Copy full SHA for e2e3a94

Commits on Sep 19, 2017

• Add an env-var TF_ENABLE_FFT_TILING_FORWARD.

  Show description for 7351390
  

  Yangzihao Wang

  authored and

  tensorflower-gardener

  committed
  7351390

  Copy full SHA for 7351390