yelihua
7c3994e48e
use Common::CreatePrefixPath instead of Common::GetRealPath
4 years ago
yelihua
801121a140
add common get realpath method
4 years ago
Zhang Qinghua
7203989c26
Support fallback feature in Graph mode (1).
4 years ago
zhouyaqiang
bc519edba2
fix compile error
4 years ago
i-robot
c928278b62
!22678 add gpu complex ops
Merge pull request !22678 from zhouyaqiang0/complex_ops
4 years ago
huanghui
8be552fc13
fix error if hccl/nccl not inited
4 years ago
zhouyaqiang
dad375abb9
add gpu complex ops
4 years ago
i-robot
66e4236bf0
!22544 fix ge backend compile problem
Merge pull request !22544 from chenmai1102/cm_master
4 years ago
i-robot
2c692bf7de
!22450 insert the overflow check operators according to the "gradients" scope name.
Merge pull request !22450 from guoqi/overflow-check-master
4 years ago
guoqi
8fccec4c20
insert overflow check operaters according to the 'gradients' scope
4 years ago
i-robot
033b4706f2
!21235 Add pass TransformOpOptimizer
Merge pull request !21235 from DeshiChen/0719_elim_transform
4 years ago
陈劢
d95160bd8a
fix ge backend compile problems
4 years ago
zengzitao
7d6c6b17bb
fix master warnings
4 years ago
dayschan
e6b1855174
Completed pass transform_op_optimizer
modifications for pass transform_op_optimizer:
1. Changed the maxflow-mincut algorithm to the Dinic's Algorithm,
since bug exists in the original ISAP codes.
if the algorithm is slow, we can apply some optimization for it. (e.g. current-arc optimization)
2. Added the pass TransformOpOptimizer in OptLevel_3.
this pass collects nodes around the specific transform operator (only Transpose now),
and use the mincut algorithm to get a plan, then re-link the original graph and
re-inference the shape and format of graph.
modifications for litegraph:
1. the class Node inherits from std::enable_shared_from_this,so we can get a shared_ptr by a pure pointer.
2. modified the Infer interface. it don't change the node, only inference the infos and return them.
4 years ago
i-robot
eba4fa8224
!22466 [GraphKernel] add fusion ops level flag for graph kernel fusion, gpu use level max, ascend use level 0
Merge pull request !22466 from r1chardf1d0/master
4 years ago
i-robot
a0ac0e2feb
!22505 Change ir save path
Merge pull request !22505 from huanghui/change-ir-save-path
4 years ago
r1chardf1d0
7ec4649805
[GraphKernel] add stable ops only flag for graph kernel fusion, and set ascend backend enable default
4 years ago
huanghui
88601da733
change ir save path to: save_graphs_path/ when alone; save_graphs_path/rank_ when distribute
4 years ago
i-robot
0ad0d9144e
!22081 [ME]Code clean of master
Merge pull request !22081 from chenfei_mindspore/code_clean_of_master
4 years ago
i-robot
16437722fd
!22465 Adam op performance optimization
Merge pull request !22465 from zhaosida/zsd_pangu
4 years ago
zhaosida
be34ccd29f
Adam op performance optimization
4 years ago
i-robot
1934e82ab2
!22399 Add errno‘s description in file_system
Merge pull request !22399 from huanghui/fix-errno
4 years ago
i-robot
32281f84e7
!19000 update LayerNormGrad split pass to V2
Merge pull request !19000 from yuchaojie/ir_fusion2
4 years ago
huanghui
7a7b1ba5c4
add errno to str in file_system
4 years ago
baihuawei
a9694a9230
ascend add nontask sink mode
4 years ago
i-robot
b58f8efa54
!22277 Clean ERROR log in PyNative Mode
Merge pull request !22277 from caifubi/master-hccl-get-rank-id
4 years ago
i-robot
abc9d8e6fe
!22068 DynamicRNN supports scenarios that hidden_size is not multiples of 16
Merge pull request !22068 from yuchaojie/op_select2
4 years ago
chenfei
72cab0262e
code clean
4 years ago
i-robot
f88445dcf0
!22260 Avoid printing meaningless error message
Merge pull request !22260 from tanghuikang/tbe_em
4 years ago
yuchaojie
1d1490df0b
xupdate LayerNormGrad split pass to V2
4 years ago
yuchaojie
7b3943bc91
DynamicRNN supports scenarios that hidden_size is not multiples of 16
4 years ago
caifubi
5210b4b353
Fix RankId/RankSize ERROR log in PyNative Mode
4 years ago
tanghuikang
9d08583307
Avoid printing meaningless error message
4 years ago
i-robot
3e7dd5ddb2
!22073 Fix codedex warning
Merge pull request !22073 from LiangZhibo/master
4 years ago
tanghuikang
81e7f7173a
Report tbe error code to ErrorManager
4 years ago
ms_yan
36a8886ca2
Revert "[feat] [assistant] [I3T96T] add new Dataset operator CMUARCTICDataset"
This reverts commit b077aa1cab .
Revert "[feat] [assistant] [I3T96X] add new Dataset operator LibriSpeechDataset"
This reverts commit 4e6f7dc97d .
delete pass_registry_test.cc
comment hiai_nlu_model_multi.pb related line
4 years ago
djc
b077aa1cab
[feat] [assistant] [I3T96T] add new Dataset operator CMUARCTICDataset
4 years ago
djc
4e6f7dc97d
[feat] [assistant] [I3T96X] add new Dataset operator LibriSpeechDataset
4 years ago
i-robot
0f3de77e93
!21149 Sync MD code clean to master
Merge pull request !21149 from xiefangqi/md_sync_codeclean_master
4 years ago
i-robot
09a1a7f1f2
!21924 Add subdirectory for glog and ir_dump
Merge pull request !21924 from huanghui/add-submodule-for-dfx-files
4 years ago
l00591931
b0cd3a735a
fix codedex error
4 years ago
huanghui
1630dcb0c8
add subdirectory for log and ir_dump
4 years ago
zhangzhaoju
9e8cbefe95
pclint clean
4 years ago
xiefangqi
0e2efc2501
fix minddata code clean
4 years ago
i-robot
0b7c4a757a
!21966 [BugFix][ops][Ascend]Adapt DynamicGruV2Grad fission v2 for RuntimeError.
Merge pull request !21966 from liuxiao93/adapter-dynamicgrv2grad-pass-v2
4 years ago
i-robot
ef535853cb
!21796 refactor TransShape
Merge pull request !21796 from yuchaojie/op_select
4 years ago
liuxiao93
bb736c108e
adapt DynamicGruV2Grad fission v2.
4 years ago
i-robot
0a04740c2c
!21572 add prelu fusion pass
Merge pull request !21572 from guyuehuo/prelu_fusion
4 years ago
i-robot
2c3bd0fdec
!21270 Add send/recv op fusion and reduce secondary stream number.
Merge pull request !21270 from linqingke/send_recv_fix
4 years ago
yuchaojie
0d56b1853c
refactor TransShape
4 years ago