diff options
author | Benjamin Kramer <kramerb@google.com> | 2018-10-09 16:52:56 -0700 |
---|---|---|
committer | TensorFlower Gardener <gardener@tensorflow.org> | 2018-10-09 17:02:47 -0700 |
commit | 65b7d0b2f84c334327a295bf41bc06c7f6b8ffe5 (patch) | |
tree | 7602939f063340ec4a98ffe5f8179fff7e3c1bd5 /tensorflow/compiler/xla/service/service.cc | |
parent | d4526cf9d1d58cbe480e7d2b8199620e0e9f0572 (diff) |
[XLA:GPU] Elide the SequentialThunk when emitting scatter with no copy
We have a 1-element thunk sequence if we're not copying. That's still two
thunks and hlo profiling gets confused if it sees two thunks for the same
instruction and one of them claims to be the whole instruction.
PiperOrigin-RevId: 216448063
Diffstat (limited to 'tensorflow/compiler/xla/service/service.cc')
0 files changed, 0 insertions, 0 deletions