You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by pa...@apache.org on 2019/10/11 14:20:30 UTC
[incubator-mxnet] branch mkldnn-v1.0 updated (bc91f5b -> 356de50)
This is an automated email from the ASF dual-hosted git repository.
patriczhao pushed a change to branch mkldnn-v1.0
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
from bc91f5b [mkldnn-v1.0] Add MKL-DNN int8 activation&pooling&flatten (#16425)
add 356de50 [mkldnn-1.0] int8 conv quantize dequantize requantize (#16283)
No new revisions were added by this update.
Summary of changes:
include/mxnet/ndarray.h | 2 +-
src/executor/graph_executor.cc | 2 +-
src/ndarray/ndarray.cc | 12 +--
src/operator/nn/mkldnn/mkldnn_base-inl.h | 15 +++
src/operator/nn/mkldnn/mkldnn_convolution-inl.h | 4 +
src/operator/quantization/dequantize.cc | 8 +-
.../quantization/mkldnn/mkldnn_dequantize-inl.h | 45 ++++-----
.../quantization/mkldnn/mkldnn_quantize-inl.h | 44 ++++-----
.../quantization/mkldnn/mkldnn_quantize_v2-inl.h | 44 ++++-----
.../quantization/mkldnn/mkldnn_quantized_conv.cc | 36 +++----
.../quantization/mkldnn/mkldnn_requantize-inl.h | 26 ++----
src/operator/quantization/quantize.cc | 6 +-
src/operator/quantization/quantize_v2.cc | 8 +-
src/operator/quantization/quantized_conv.cc | 4 +-
src/operator/quantization/requantize.cc | 6 +-
src/operator/subgraph/mkldnn/mkldnn_conv-inl.h | 6 +-
src/operator/subgraph/mkldnn/mkldnn_conv.cc | 103 ++++++++++-----------
.../subgraph/mkldnn/mkldnn_conv_property.h | 2 +-
.../mkldnn/mkldnn_post_quantize_property.h | 2 +-
.../subgraph/mkldnn/mkldnn_subgraph_base-inl.h | 2 +-
.../subgraph/mkldnn/mkldnn_subgraph_property.cc | 22 +++--
21 files changed, 198 insertions(+), 201 deletions(-)