提交 · f649442dddefbd69b057174cd731d2d0fee18a29 · Crayon鑫 / Paddle

10 2月, 2021 1 次提交

New custom operator extension mechanism (#30690) · f649442d

由 Chen Weihang 提交于 2月 09, 2021

* initial commit: simple demo

* polish copyright format

* add grap op simple demo

* adapt uncertain number of argument

* change trait marco name

* add place & dtype support for add kernel

* add dispath and infershape func

* poish code & add notes

* add dynamic_loader dep for paddle_framework

* add new custom op test dir

* polish impl details

* add unittest for new custom op

* fix failed unittest

* Costum op (#1)

* fix compile error

* wrap framework tensor with LoDTensor

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* add CustomTensor default constructor

* add size() for CustomTensor

* make size const for CustomTensor

* refactor place related api to circle the concept

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* make place const

* make Tensor copy

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* remove additional head of framework

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* add gpu test

* merge latest cwh code in

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* Remove ShareData from user && Change CustomTensor to Tensor && Support more data type (#2)

* fix compile error

* wrap framework tensor with LoDTensor

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* add CustomTensor default constructor

* add size() for CustomTensor

* make size const for CustomTensor

* refactor place related api to circle the concept

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* make place const

* make Tensor copy

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* remove additional head of framework

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* add gpu test

* merge latest cwh code in

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* hid share data from and to

* rename CustomTensor to Tensor

* refactor register design & add test

* change op_funtion to op_meta_info

* split op meta info into .h and .cc

* move get methods into friend class

* move OpMetaInfoHelper into framework space

* move CustomTensorUtils into framework space

* change pybind api name

* move PD C API into op meta info

* add register custom op api

* remove inference cmake change

* refactor copy to api && change Reshape to lowercase && support more dtype && add more test (#3)

* fix compile error

* wrap framework tensor with LoDTensor

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* add CustomTensor default constructor

* add size() for CustomTensor

* make size const for CustomTensor

* refactor place related api to circle the concept

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* make place const

* make Tensor copy

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* remove additional head of framework

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* add gpu test

* merge latest cwh code in

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* hid share data from and to

* rename CustomTensor to Tensor

* support multi dtype

* remove lod, make reshape lowercase, add copy test and refactor copy api

* remove lod, make reshape lowercase, add copy test and refactor copy api

* remove lod, make reshape lowercase, add copy test and refactor copy api

* remove lod, make reshape lowercase, add copy test and refactor copy api

* fix copy to error

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* polish detail & error message

* polish test details

* Add cast api && Change copy related api to copy_to && add more test (#4)

* fix compile error

* wrap framework tensor with LoDTensor

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* add CustomTensor default constructor

* add size() for CustomTensor

* make size const for CustomTensor

* refactor place related api to circle the concept

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* fix compile error

* make place const

* make Tensor copy

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* debug CustomTensor core

* remove additional head of framework

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* use back to shared ptr for custom tensor

* add gpu test

* merge latest cwh code in

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* adjust ut code of custom op

* hid share data from and to

* rename CustomTensor to Tensor

* support multi dtype

* remove lod, make reshape lowercase, add copy test and refactor copy api

* remove lod, make reshape lowercase, add copy test and refactor copy api

* remove lod, make reshape lowercase, add copy test and refactor copy api

* remove lod, make reshape lowercase, add copy test and refactor copy api

* fix copy to error

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add more test

* add type cast

* add cast and make copy to api

* add cast and make copy to api

* add cast and make copy to api

* add cast and make copy to api

* merge cwh code

* merge cwh code

* merge cwh code

* merge cwh code

* merge cwh code

* add more error log

* add more error log

* polish code

* used for test

* remove test comment

* remove test comment

* fix uint8 type error

* fix lost uint8 type error

* add test for coverage

* polish details by reviewer comments

* add prefix for DISABLE_COPY_AND_ASSIGN
Co-authored-by: NJiabin Yang <360788950@qq.com>

f649442d

05 2月, 2021 1 次提交

Performance optimization for dynamic setitem: Call op set_value to speed up... · 39f41cb4

由 liym27 提交于 2月 05, 2021

Performance optimization for dynamic setitem: Call op set_value to speed up because the original call to TensorToPyArray will introduce unnecessary data copy. (#30817)

39f41cb4

04 2月, 2021 2 次提交
- J
  Add bf16 fast performance verification (#30551) · 73cdea01
  由 joanna.wozna.intel 提交于 2月 04, 2021
```
* Update Xbyak and add bf16 fast performance verification

* Fix formating

* Change LOG message

* Trigger an update of a new tag
```
  73cdea01
- W
  
  fix xpu dygraph place (#30868) · 6e3856d3
  由 WangXi 提交于 2月 04, 2021
  
  6e3856d3
03 2月, 2021 2 次提交
- 石
  support xpu with analysis predictor, test=develop (#30832) · 2ac4143b
  由石晓伟提交于 2月 03, 2021
```
* support xpu inference with analysis predictor, test=develop

* merge the cmake of the xpu toolchain, test=develop

* add c-apis, test=develop

* fix a bug in extern_xpu, test=develop
```
  2ac4143b
- W
  
  【kunlun】dygraph supports multi xpu card training (#30671) · b1026f64
  由 WangXi 提交于 2月 03, 2021
  
  b1026f64
01 2月, 2021 1 次提交
- T
  dump to cpu (#30750) · cb66c53c
  由 Thunderbrook 提交于 2月 01, 2021
```
* dump to cpu

* format

* format

* format
```
  cb66c53c
29 1月, 2021 1 次提交
- S
  
  rm Singleton of reducer (#30775) · 3858f458
  由 ShenLiang 提交于 1月 29, 2021
  
  3858f458
25 1月, 2021 1 次提交

add DLA support：C++&&Python api (#30165) · ae0f88a9

由 Shang Zhizhou 提交于 1月 25, 2021

* add dla

* add dla done

* add python api
Co-authored-by: Nshangzhizhou <root@szth-rp-fanyi-opera49.szth.baidu.com>

ae0f88a9

21 1月, 2021 1 次提交
- T
  solve build gpu task core (#30626) · 1bebc092
  由 Thunderbrook 提交于 1月 21, 2021
```
* build gpu task core

* format
```
  1bebc092
20 1月, 2021 2 次提交

use nvtx push pop in timeline (#30567) · 90773473

由 wanghuancoder 提交于 1月 20, 2021

* delete empty line of pybing.cc, test=develop

* use nvtx push pop in timeline, test=develop

* change year, test=develop

* add #ifdef PADDLE_WITH_CUDA, test=develop

* add #ifndef WIN32, test=develop

* is_pushed to is_pushed_, test=develop

90773473

add some RecordEvent, for dygraph timeline (#30299) · d1b25ed9

由 wanghuancoder 提交于 1月 20, 2021

* add some RecordEvent, for dygraph timeline, test=develop

* change GpuMemcpySync to memory::Copy, test=develop

* fix compile problem, test=develop

* fix compile problem, test=develop

* fix, test=develop

* fix, test=develop

d1b25ed9

19 1月, 2021 2 次提交
- L
  support layer_norm fp16 in dygraph amp (#30430) · 7043b8cf
  由 Leo Chen 提交于 1月 19, 2021
```
* support layer_norm fp16 in dygraph amp

* add ut

* refine code
```
  7043b8cf
- W
  
  delete empty line of pybing.cc, test=develop (#30529) · 59ad6ff3
  由 wanghuancoder 提交于 1月 19, 2021
  
  59ad6ff3
18 1月, 2021 2 次提交
- H
  
  Ascend Framework Part2: pybind files (#30410) · e207fe63
  由 hutuxian 提交于 1月 18, 2021
  
  e207fe63
- W
  
  if pybind.cc changed, generate total report, test=develop (#30514) · bd971922
  由 wanghuancoder 提交于 1月 18, 2021
  
  bd971922
17 1月, 2021 1 次提交
- G
  Modify the calculation logic of LambOptimizer (#29313) · 11e78eba
  由 guofei 提交于 1月 17, 2021
```
* Modify the calculation logic of LambOptimizer
```
  11e78eba
15 1月, 2021 1 次提交

Add Inplace strategy (Output reuse Input Varbase) in dygraph (#30103) · 13d75736

由 pangyoki 提交于 1月 15, 2021

* add view strategy on squeeze,unsqueeze,reshape,flatten

* add squeeze unittest

* add unittests

* use View strategy as name rather than Reuse Allacation

* fix view api doc

* fix format

* use core.ops when input of reshape2 is Tensor

* fix test_cross_entropy_loss error because of reshape2

* fix test_cross_entropy_loss error because of reshape2

* add inplace strategy

* add elementwise_add sub

* let backward op not use inplace

* grad op do not use inplace

* fix memory increase error and add leaf error message

* delete selected_rows

* change op_function

* little change

* solve HandleViewBetweenInputAndOutput

* add unittest and leaf error message

* merge view error

* optimize op_function_generator format and support sum inplace op

* fix format of basic_engine

* fix format for framework

* little change of variable wrapper

* add reshape, squeeze, unsqueeze, scatter api

* add relu elu tanh softmax inplace api

* fix test_squeeze_op unittest

* fix test_relu_op unittest

* fix comment problems

* delete sample code of inplace api

* add reference of grad_pending_nodes in basic_engine

* fix unittest name

* add inplace apis into wlist

* fix error message

* add PADDLE_ENFORCE for set grad op twice

* fix head file error

13d75736

14 1月, 2021 1 次提交
- Y
  
  Heter ps new (#30198) · 6e0da01c
  由 yaoxuefeng 提交于 1月 14, 2021
  
  6e0da01c
13 1月, 2021 2 次提交

Set expected place in child thread for dataloader to avoid costing cuda memory... · 3d015f1c

由 Leo Chen 提交于 1月 13, 2021

Set expected place in child thread for dataloader to avoid costing cuda memory on other card (#30338)

* set expected place in child thread for dataloader

* set device id when set tensor from numpy

* revert tensor_py change

* add compile guard

* fix ci

* fix bug

3d015f1c

S

Support unused parameters in dynamic graph distributed (#30224) · a60f17b8
由 ShenLiang 提交于 1月 13, 2021

a60f17b8

12 1月, 2021 2 次提交
- T
  Fix/distributed proto (#29981) · 25f80fd3
  由 tangwei12 提交于 1月 12, 2021
```
* rename sendrecv.proto to namespace paddle.distributed

* split ps with distributed
```
  25f80fd3
- C
  【Paddle.Fleet】Support local save sparse param (#30175) · d479ae17
  由 Chengmo 提交于 1月 12, 2021
```
* add save tensor support
Co-authored-by: NseiriosPlus <tangwei12@baidu.com>
```
  d479ae17
11 1月, 2021 1 次提交
- A
  
  Add tf32 switch for cuDNN (#29192) · 924aac22
  由 AshburnLee 提交于 1月 11, 2021
  
  924aac22
09 1月, 2021 1 次提交

add View(reuse allocation) strategy on squeeze, unsqueeze, reshape, flatten op (#29913) · da16b33f

由 pangyoki 提交于 1月 09, 2021

* add view strategy on squeeze,unsqueeze,reshape,flatten

* add squeeze unittest

* add unittests

* use View strategy as name rather than Reuse Allacation

* fix view api doc

* fix format

* use core.ops when input of reshape2 is Tensor

* fix test_cross_entropy_loss error because of reshape2

* delete selected_rows

* change op_function

* little change

* solve HandleViewBetweenInputAndOutput

da16b33f

08 1月, 2021 3 次提交

Fix dtype of ungenerated grad var (#28511) · 8696335f

由 Leo Chen 提交于 1月 08, 2021

* fix dtype of ungenerated grad var

* update ut

* refine code

* set default dtype

* fix could_use_cudnn bug

* remove debug code

* re-implement

* fix bug

8696335f

Add callback after TensorCopy (#30123) · 1f97d61c

由 Leo Chen 提交于 1月 08, 2021

* change to tensor copy sync

* change to tensor copy sync

* make copy_to safe when use TensorCopy

* refine code

* add ut

* add cudapinned garbagecollector

* add testcase: cpu place -> cuda pinned place

1f97d61c

C
【Paddle.Fleet】Fix tensor table (#30075) · 528e03fc
由 Chengmo 提交于 1月 08, 2021
```
* add tensor table
```
528e03fc

07 1月, 2021 1 次提交
- 1
  Add Lookahead and ModelAverage Optimizer (#30004) · 198fbdfb
  由 123malin 提交于 1月 07, 2021
```
* test=develop, add model_average and lookahead
```
  198fbdfb
06 1月, 2021 2 次提交

add dispenable input for core.ops.reshape2/expand/slice (#30072) · adac38c5

由 Leo Chen 提交于 1月 06, 2021

* add dispenable input 'shape' for core.ops.reshape2

* add dispenable inputs for core.ops.reshape2/expand/slice

* add ut

adac38c5

Fix bug: In dynamic mode, if start or end is negetive, __getitem__ return wrong result(#30003) · 9922bd41

由 liym27 提交于 1月 06, 2021

1. when slice_item is a slice: 
 1) the start of __getitem__ should be std::max(start, 0) if slice
 2) the start of __getitem__ should be std::min(end, dim) 
2. when slice_item is an integer, it should be in [-dim_len, dim_len) 
3. Fix error message to use accurate data

9922bd41

05 1月, 2021 1 次提交
- T
  add topo-aware in heter-ps (#30087) · 0b8e1fad
  由 Thunderbrook 提交于 1月 05, 2021
```
* add topo aware

* resource.h

* topo aware

* format
```
  0b8e1fad
04 1月, 2021 1 次提交
- C
  [Inference] zero_copy_tensor supports int8_t (#30053) · 68398abc
  由 cc 提交于 1月 04, 2021
```
* zero_copy_tensor supports int8_t
```
  68398abc
27 12月, 2020 1 次提交

[Dynamic Inplace] Support ShareInplaceVersionCounterWith for C++ Tensor (#29842) · 9602a182

由 liym27 提交于 12月 27, 2020

* Revert "[inplace] Add ShareHolderWith for class Variable and SharePlaceholderWith in VarBase.detach() to share the same Tensor/SelectedRows (#29267)"

This reverts commit b10ecd9d.

* Support ShareInplaceVersionCounterWith to share the same inplace version counter for VarBase

9602a182

26 12月, 2020 1 次提交
- L
  
  [Kunlun] PR2: Support MultiDevicePass and BKCL in parallel executor (#29574) · 4427df37
  由 liuyuhui 提交于 12月 26, 2020
  
  4427df37
24 12月, 2020 1 次提交

[Feature] one ps (3/4) (#29604) · 032414ca

由 tangwei12 提交于 12月 24, 2020

* oneps (3/4)
Co-authored-by: NMrChengmo <cmchengmo@163.com>
Co-authored-by: Nmalin10 <malin10@baidu.com>
Co-authored-by: Nchengmo <chengmo@baidu.com>

032414ca

23 12月, 2020 1 次提交

heter box (#29734) · 09b6e719

由 Thunderbrook 提交于 12月 23, 2020

* 　add heter box

* add trainer, worker, wrapper...

* format

* for ci

* format

* remove boost get

* boost & copyright

* rename

* 　rename

* format

* format

* format
Co-authored-by: Nyaoxuefeng6 <yaoxuefeng@baidu.com>

09b6e719

22 12月, 2020 1 次提交
- S
  Support multi-stream communication for dynamic graph distributed (#29525) · 01e2874a
  由 ShenLiang 提交于 12月 22, 2020
```
* fix fleet for multi-stream

* fix memcpy for ncclid

* use sync to solve move operation
```
  01e2874a
16 12月, 2020 2 次提交

L

[Kunlun] PR1:Support one Kunlun card training in parallel executor (#29337) · f13c3a9c
由 liuyuhui 提交于 12月 16, 2020

f13c3a9c

添加rocm平台支持代码 (#29342) · 76738504

由 Y_Xuan 提交于 12月 16, 2020

* 添加rocm平台支持代码

* 修改一些问题

* 修改一些歧义并添加备注

* 修改代码格式

* 解决冲突后的代码修改

* 修改operators.cmake

* 修改格式

* 修正错误

* 统一接口

* 修改日期

76738504

Crayon鑫 / Paddle 与 Fork 源项目一致

Crayon鑫 / Paddle
与 Fork 源项目一致