This is an automated email from the ASF dual-hosted git repository.
haibin pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
from 521e925 [Numpy] Numpy behavior random.uniform() (#15858)
add bdeb7bc Softmax optimization for GPU (#15545)
No new revisions were added by this update.
Summary of changes:
src/common/cuda_utils.cc | 90 +++
src/common/cuda_utils.h | 50 +-
src/operator/mxnet_op.h | 30 +
src/operator/nn/log_softmax.cc | 78 +++
.../{tensor/cast_storage.cu => nn/log_softmax.cu} | 16 +-
src/operator/nn/softmax-inl.h | 723 ++++++++++++---------
src/operator/nn/softmax.cc | 99 ---
src/operator/nn/softmax.cu | 18 +-
src/operator/nn/softmin.cc | 89 +++
.../{tensor/cast_storage.cu => nn/softmin.cu} | 16 +-
10 files changed, 760 insertions(+), 449 deletions(-)
create mode 100644 src/common/cuda_utils.cc
create mode 100644 src/operator/nn/log_softmax.cc
copy src/operator/{tensor/cast_storage.cu => nn/log_softmax.cu} (67%)
create mode 100644 src/operator/nn/softmin.cc
copy src/operator/{tensor/cast_storage.cu => nn/softmin.cu} (67%)
