mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
1eed54d17a
Summary: Bump oneDNN (mkl-dnn) to 1.7 for bug fixes and performance optimizations - Fixes https://github.com/pytorch/pytorch/issues/42115. Fixed build issue on Windows for the case when oneDNN is built as submodule - Fixes https://github.com/pytorch/pytorch/issues/45746. Fixed segmentation fault for convolution weight gradient on systems with Intel AVX512 support This PR also contains a few changes in ideep for follow-up update (not enabled in current PR yet): - Performance improvements for the CPU path of Convolution - Channel-last support Pull Request resolved: https://github.com/pytorch/pytorch/pull/47853 Reviewed By: bdhirsh Differential Revision: D25275268 Pulled By: VitalyFedyunin fbshipit-source-id: 75a589d57e3d19a7f23272a67045ad7494f1bdbe
49 lines
2.2 KiB
C++
49 lines
2.2 KiB
C++
#pragma once
|
|
|
|
#include <caffe2/core/macros.h> // For caffe2 macros.
|
|
#include <caffe2/utils/eigen_utils.h>
|
|
// All caffe2 ideep related headers
|
|
#include <ideep.hpp>
|
|
#include <caffe2/ideep/utils/ideep_context.h>
|
|
#include <caffe2/ideep/utils/ideep_operator.h>
|
|
|
|
namespace caffe2 {
|
|
|
|
enum ConvAlgorithm {
|
|
CONV_ALGORITHM_AUTO = 0,
|
|
CONV_ALGORITHM_WINOGRAD = 1,
|
|
CONV_ALGORITHM_MAX
|
|
};
|
|
|
|
enum FusionType {
|
|
FUSION_UNKNOWN = 0,
|
|
FUSION_CONV_RELU = 1,
|
|
FUSION_CONV_SUM = 2,
|
|
FUSION_CONV_SUM_RELU = 3,
|
|
FUSION_MAX
|
|
};
|
|
|
|
#define USE_IDEEP_DEF_ALIASES() \
|
|
/* the tensor type created/handled by iDEEP */ \
|
|
using itensor = ideep::tensor; \
|
|
/* the date layout of iDEEP tensor */ \
|
|
using iformat = ideep::format_tag; \
|
|
/* the scales for iDEEP tensor with different data type */ \
|
|
using iscale = ideep::scale_t; \
|
|
/* the detial algorithm for iDEEP operators, e.g. winograd */ \
|
|
using ialgo = ideep::algorithm; \
|
|
/* the kind of propagation for iDEEP operators, e.g. forward, training */ \
|
|
using iprop = ideep::prop_kind; \
|
|
/* the kind of low precision operators, e.g. signed/unsigned activation */ \
|
|
using ilowp_kind = ideep::lowp_kind; \
|
|
/* the data type of iDEEP tensor, e.g. f32, u8, s8 */ \
|
|
using idtype = ideep::tensor::data_type; \
|
|
/* the descriptor of iDEEP tensor */ \
|
|
using itdesc = ideep::tensor::descriptor; \
|
|
/* the attribute for operator to describe the details of inputs&fusion */ \
|
|
using iattr = ideep::attr_t; \
|
|
/* the detail flags for batch normalization */ \
|
|
using ibn_flag = ideep::batch_normalization_flag;
|
|
|
|
} // namespace caffe2
|