- 24 9月, 2020 1 次提交
-
-
由 Mihai Maruseac 提交于
-
- 24 6月, 2020 1 次提交
-
-
由 Deven Desai 提交于
Starting with ROCm 3.5 the underlying compiler used by hipcc will change from HCC to hip-clang. There will be a corresponding change in the HIP Runtime as well. This commit is part of a series which are intended to make the transition to ROCm 3.5+ easier. Disabling some unit-tests (by adding the no_rocm tag to them) because they regerss after the switch to ROCm 3.5+ ``` # XLA related failure //tensorflow/python/eager:def_function_xla_jit_test //tensorflow/python/keras:backend_test # eigen packet then scalar JIRA ticket 236756 //tensorflow/python/ops/ragged:ragged_dispatch_test # genuine miscompares //tensorflow/python/kernel_tests:bias_op_deterministic_test //tensorflow/python/kernel_tests:array_ops_test # Memory access fault by GPU ... Reason: Page not present or supervisor privilege. //tensorflow/python/ops/parallel_for:math_test ```
-
- 24 9月, 2019 1 次提交
-
-
由 Tres Popp 提交于
PiperOrigin-RevId: 270858131
-
- 28 8月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 265883639
-
- 12 6月, 2019 1 次提交
-
-
由 Yanan Cao 提交于
PiperOrigin-RevId: 252724110
-
- 11 6月, 2019 2 次提交
-
-
由 A. Unique TensorFlower 提交于
Adjust structure of all BUILD files to recommended style (https://docs.bazel.build/versions/master/skylark/build-style.html#file-structure), moving loads to top. PiperOrigin-RevId: 252573670
-
由 Scott Zhu 提交于
PiperOrigin-RevId: 252440492
-
- 07 6月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 251943109
-
- 24 5月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 249807024
-
- 09 5月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 247275924
-
- 18 4月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
Support broadcasting BatchMatMulV2 in parallel_for. PiperOrigin-RevId: 244106976
-
- 14 3月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
I ran each test 200 times, and these passed without any flakes. PiperOrigin-RevId: 238414725
-
- 14 2月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
One test does not test floor_div when running on XLA due to XLA's use of an Nvidia speed optimization for approximate division. This is a no-op for Tensorflow Classic. PiperOrigin-RevId: 233827805
-
- 13 2月, 2019 1 次提交
-
-
由 Jeffrey Poznanovic 提交于
-
- 30 1月, 2019 1 次提交
-
-
由 Gaurav Jain 提交于
PiperOrigin-RevId: 231529445
-
- 23 1月, 2019 1 次提交
-
-
由 Anna R 提交于
PiperOrigin-RevId: 230396943
-
- 14 1月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
Some test methods are disabled, but all tests now have a new "_xla" version of the test for XLA:GPU testing. This will run 2 different tests. One with XLA and one without. PiperOrigin-RevId: 229149574
-
- 10 1月, 2019 1 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 228620325
-
- 19 9月, 2018 1 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 213610324
-
- 11 8月, 2018 1 次提交
-
-
由 Rachel Lim 提交于
PiperOrigin-RevId: 208255119
-
- 17 7月, 2018 1 次提交
-
-
由 A. Unique TensorFlower 提交于
precision of fused winograd implementation of convolution backprop. PiperOrigin-RevId: 204803121
-
- 03 7月, 2018 1 次提交
-
-
由 A. Unique TensorFlower 提交于
enables applications like auto-batching, jacobians, per-example gradients. PiperOrigin-RevId: 203026617
-