projects / platform / upstream / tensorflow.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a5e1809) | patch
[XLA:GPU] Zero out input buffers before running cudnn conv autotune.
authorJustin Lebar <jlebar@google.com>
Sat, 5 May 2018 19:34:32 +0000 (12:34 -0700)
committerTensorFlower Gardener <gardener@tensorflow.org>
Mon, 7 May 2018 23:21:52 +0000 (16:21 -0700)
commit094daf9d005ffe9322602988dbff033c3e359cef
tree36b561450270b74aa5fa9d85a8137a1ef4782a3dtree | snapshot
parenta5e1809ddedbff8d9f66b82f9cf0989976694050commit | diff
[XLA:GPU] Zero out input buffers before running cudnn conv autotune.

We don't need a corresponding change in gemm_thunk.cc because for gemms,
we do our autotune at runtime, at which point we have some real data in
our input/output buffers.

PiperOrigin-RevId: 195548896
tensorflow/compiler/xla/service/gpu/cudnn_convolution_algorithm_picker.cc diff | blob | history
Domain: Machine Learning / ML Framework;