提交 · 07514139347328aff3b01f19301dac2ab601ea06 · BaiXuePrincess / Paddle

16 1月, 2023 1 次提交

add gpu_cpu_map_matmul_to_mul_pass to kGpuLowerPrecisionPasses (#49753) · 07514139

由 Yuanle Liu 提交于 1月 16, 2023

* add gpu_cpu_map_matmul_to_mul_pass to kGpuLowerPrecisionPasses

* disable fc_elementwise_layernorm_fuse_pass in mixed precision

07514139

13 1月, 2023 1 次提交
- W
  add oss flash fmha and fmhca support (#49438) · a48b8e2c
  由 Wang Bojun 提交于 1月 13, 2023
```
* add fmha_flashattention oss plugin
```
  a48b8e2c
11 1月, 2023 1 次提交
- Z
  fix paddle_infer_contrib inclue (#49720) · 24f5c46e
  由 zhangxin81 提交于 1月 11, 2023
```
* fix paddle_infer_contrib include
```
  24f5c46e
10 1月, 2023 2 次提交
- X
  
  add_paddle_test (#49640) · f4d267c2
  由 xiaoxiaohehe001 提交于 1月 10, 2023
  
  f4d267c2
- S
  
  Add reduce_min prod trt converter (#49615) · 13992de7
  由 Sanbu 提交于 1月 10, 2023
  
  13992de7
09 1月, 2023 2 次提交
- W
  Preln groupnorm (#49463) · 591be3bd
  由 wenbin 提交于 1月 09, 2023
```
* skip_groupnorm

* init

* preln

* add ut

* more assert

* set timeout

* fix windows ci issue
```
  591be3bd
- G
  
  Unify the pass of the map class (#49568) · ee49994f
  由 gem5 提交于 1月 09, 2023
  
  ee49994f
06 1月, 2023 2 次提交
- Y
  
  [Inference] fix pass_builder (#49595) · 44cb3da3
  由 Yuanle Liu 提交于 1月 06, 2023
  
  44cb3da3
- Y
  
  fix trt engine memory sharing (#49584) · 1e8976e8
  由 Yuanle Liu 提交于 1月 06, 2023
  
  1e8976e8
05 1月, 2023 2 次提交
- W
  
  [Inference] inplace all reshape op (#49146) · 017af746
  由 Wilber 提交于 1月 05, 2023
  
  017af746
- Y
  
  [Paddle Inference] add unitest for zero_copy_tensor with bool type (#49495) · 8705a79d
  由 Yuanle Liu 提交于 1月 05, 2023
  
  8705a79d
04 1月, 2023 1 次提交
- L
  
  add multi_devices_fused_multi_transformer_encoder_pass and cherry-pick from 48349 (#49383) · 29eec2dd
  由 lzy 提交于 1月 04, 2023
  
  29eec2dd
03 1月, 2023 3 次提交
- Y
  
  [Paddle Inference] enhance paddle_infer::Tensor data type (#49388) · dc13f7c5
  由 Yuanle Liu 提交于 1月 03, 2023
  
  dc13f7c5
- Z
  [Paddle Inference] Implement conv2d_fusion NHWC format using cutlass (#47989) · c123dd1e
  由 zhoutianzi666 提交于 1月 03, 2023
```
* Implement conv2d_fusion NHWC format using CUTLASS
* Add unit testing for CUTLASS Conv in inference
* Add experimental API for CUTLASS.
```
  c123dd1e
- S
  
  Add not_equal trt converter (#49393) · 822ea0f9
  由 Sanbu 提交于 1月 03, 2023
  
  822ea0f9
28 12月, 2022 1 次提交
- Y
  
  update some trt log (#49330) · 02019804
  由 Yuanle Liu 提交于 12月 28, 2022
  
  02019804
22 12月, 2022 1 次提交
- G
  
  Enable identity_scale_op_clean_pass by default (#49227) · 9dac1e71
  由 gem5 提交于 12月 22, 2022
  
  9dac1e71
21 12月, 2022 1 次提交

Refactor Pass for fused_conv (#48848) · 7f0eb2e3

由 zyfncg 提交于 12月 21, 2022

* refactor conv_activation_mkldnn_fuse_pass

* refactor conv_affine_channel_mkldnn_fuse_pass

* fix conv_activation_mkldnn_fuse_pass

* fix mkldnn unittest

* refactor int8_scale_calculation_mkldnn_pass and params_quantization_mkldnn_pass

* refactor conv_elementwise_add_mkldnn_fuse_pass

* fix quant

* refactor conv_bn_fuse_pass

* fix conv_bn_fuse_pass

* refactor depthwise_conv_bn_fuse_pass

* fix unittest

* fix conv_bn_fuse_pass

* remove redundant conv2d in params_quantization_mkldnn_pass

* fix params_quantization_mkldnn_pass_tester

7f0eb2e3

20 12月, 2022 2 次提交
- X
  
  fix_arguments (#49186) · b0e9e48d
  由 xiaoxiaohehe001 提交于 12月 20, 2022
  
  b0e9e48d
- R
  
  [Paddle Inference] Add add arg_min trt converter (#49113) · 44973c65
  由 Ryan 提交于 12月 20, 2022
  
  44973c65
19 12月, 2022 1 次提交
- W
  [Paddle Inference] General optimization for no_varlen skiplayernorm (#49039) · b50dbe0b
  由 Wangzheee 提交于 12月 19, 2022
```
* General optimization for no_varlen embedding layernorm
```
  b50dbe0b
17 12月, 2022 1 次提交
- X
  
  [Paddle Inference] Memory Optimize destruct argument (#49046) · 0b36655b
  由 xiaoxiaohehe001 提交于 12月 17, 2022
  
  0b36655b
16 12月, 2022 1 次提交
- Y
  
  add PADDLE_WITH_TENSORRT to EnableTensorRtEngine (#49091) · c4f30c51
  由 Yuanle Liu 提交于 12月 16, 2022
  
  c4f30c51
15 12月, 2022 4 次提交

Z
[inference] move IsFloatVar() from tensorrt/ to api/ (#49070) · 2190ea09
由 Zhang Jun 提交于 12月 15, 2022
```
* move IsFloatVar() from tensorrt/ to api/
```
2190ea09
H

[PHI decoupling] move softmax from fluid to phi and remove cpu_vec.h in fluid (#48970) · 344b99e1
由 huangjiyi 提交于 12月 15, 2022

344b99e1

[PHI decoupling] Remove fluid imports from MKLDNN code (#48981) · 4d5a5533

由 Sławomir Siwek 提交于 12月 15, 2022

* fix wrong handler name

* mkldnn_engine -> onednn_engine

* remove fluid/errors.h imports

* remove fluid/enforce.h imports

* remove note and unnecessary import

* remove fluid/pretty_log.h imports

* remove fluid/place.h imports

* remove fluid/data_layout_transform.h imports

* remove fluid/device_context.h imports

* remove mkldnn_helper code

* remove fluid/mkldnn_reuse.h imports

* pretty_log import

4d5a5533

W
[Inference] memory_optimize and mkdlnn problem (#49054) · 04dd2861
由 Wilber 提交于 12月 15, 2022
```
* memory_optimize and mkdlnn problem

* update

* update

* update
```
04dd2861

14 12月, 2022 3 次提交
- Y
  
  [Paddle Inference] rewrite convert_to_mixed_precision (#48853) · 28ea9aad
  由 Yuanle Liu 提交于 12月 14, 2022
  
  28ea9aad
- H
  Deleted mkldnn_inplace_pass code (#47818) · 3cfb2e1a
  由 Hulek 提交于 12月 14, 2022
```
* Deleted mkldnn_inplace_pass code

* Fixed error with cmake

* Resolve conflicts
```
  3cfb2e1a
- Z
  [inference][trt] add more unary op and square (#48534) · e6cabea1
  由 Zhang Jun 提交于 12月 14, 2022
```
* add more unary op and square
```
  e6cabea1
13 12月, 2022 1 次提交
- E
  
  enable custom device save model on device memory && fix conflict (#48221) · b6aa9f53
  由 engineer1109 提交于 12月 13, 2022
  
  b6aa9f53
12 12月, 2022 1 次提交
- F
  
  fix: Move the pass location to the appropriate location (#48951) · 6698e8d1
  由 feng_shuai 提交于 12月 12, 2022
  
  6698e8d1
11 12月, 2022 1 次提交
- W
  
  fix for mkldnn (#48852) · 96e58f87
  由 Wilber 提交于 12月 11, 2022
  
  96e58f87
09 12月, 2022 2 次提交
- Y
  [Inference] optimize some code and fix some bug (#48780) · c0034b5b
  由 Yuanle Liu 提交于 12月 09, 2022
```
* clean ir_pass_manager and fix map_depthwise_conv_to_conv_pass

* fix unitest timeout
```
  c0034b5b
- P
  
  [PHI decoupling] move "flags.h" from fluid to phi (#48696) · 39ffef0d
  由 PuQing 提交于 12月 09, 2022
  
  39ffef0d
08 12月, 2022 5 次提交
- R
  rewrite delete_weight_dequant_linear_op_encoder/decoder pass (#48650) · 95332bef
  由 RichardWooSJTU 提交于 12月 08, 2022
```
* rewrite delete_weight_deqquant_linear_op_encoder/decoder pass
```
  95332bef
- W
  [Paddle Inference] General optimization for no_varlen embedding layernorm (#48580) · 22bfa579
  由 Wangzheee 提交于 12月 08, 2022
```
* general optimization no_varlen embedding layernorm
```
  22bfa579
- W
  
  [Inference] Enable infer shape cache. (#48312) · f88713e1
  由 Wilber 提交于 12月 08, 2022
  
  f88713e1
- 六
  [Paddle Inference] Add add onehot trt converter (#48655) · 1adf5430
  由六个骨头提交于 12月 08, 2022
```
* add onehot trt converter

* add unitest

* fix bug

* opt code

* fix bug

* fix depth_tensor

* fix unitest

* fix bug

* fix unitest

* fix bug

* fix bug

* fix bug

* fix bug
```
  1adf5430
- W
  
  [Inference] inference add cinn interface (#48741) · 3a387df6
  由 Wilber 提交于 12月 08, 2022
  
  3a387df6

BaiXuePrincess / Paddle 与 Fork 源项目一致

BaiXuePrincess / Paddle
与 Fork 源项目一致