p20518467
/
InfiniTensor
forked from jiuyuan/InfiniTensor
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
159 Commits 70 Branches 0 Tags 141 MiB
Commit Graph

94 Commits

Author SHA1 Message Date
bolun 7945693131 Add: benchmark for cudnn softmax 2023-07-20 11:03:07 +08:00
constroy 579cdbbb81
fix ReduceMean and element_wise (#90) 
* feat: 导出 getPerfTime 到 python

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix parsing of ReduceMean

* ReduceMean axes defaults to None

* fix ElementWiseCudnn with shape broadcasting

* fix format

---------

Signed-off-by: YdrMaster <ydrml@hotmail.com>
Co-authored-by: YdrMaster <ydrml@hotmail.com>
 2023-06-29 07:15:07 +08:00
YdrMaster 26f0d13c26
Dev for 202303ddl (#66) 
* add activation operatiopn relu, tanh, sigmoid on mlu

* commit for format

* add activation backward operation

* add test for activation_backward

* add test

* add convbpfilter

* fix

* add transpsoe code and test

* add trigon function operation on mlu: sin,cos,tan,asin,sinh,asinh

* add copy operation on mlu

* add ceil operation and floor operation

* add operation clip

* add operation cnnl div, test and test for divdemo bangc kernel

* add divnonan operation and test

* add erf operation

* add exp operation

* add operation fill

* add log operation

* add log1p operation

* add l2loss operation

* add maximum and minimum operation

* add mseloss operation

* add negTensor operation

* add power operation

* add reciprocal operation

* add sqrt and rsqrt operation

* add transform operation

* add addn operation

* add muln operation

* cherrry pick some operation

* add floordiv operation and floordivtrunc operation

* add floormod operation

* add cumsum operation

* add det operation

* add pad operation

* format

* add concat operation

* format

* add split operation

* fix concat and split operation

* add round operation

* add pooling operation

* add square operation

* add squaredDifference operation

* code format fix

* add flip operation

* code format fix

* add hardtanh operation

* add logic operation

* add addcdiv and addcmul operation

* add arange operation

* add bitcompute operation

* add net test

* fmt

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* style: rename

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: 用 NativeCpuRuntime 替换 CpuRuntime

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix code

* fix code

* fix code by review suggestion

* remove operation which is not the onnx operation

* fix format

* clang format

* refactor: tensor 的 print 加一层模板的 dataToString

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: onnx 导出

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 增加计算图优化接口

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* add clip operation

* feat: 支持导入 clip

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* test: 导入导出测试加入 ci

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix batch norm

* feat: 增加 Shape 算子

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 支持导入 unsqueeze

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: 修正 clip 接口

feat: 支持导入 transpose
Signed-off-by: YdrMaster <ydrml@hotmail.com>

* add broadcast operation

* fix elementwise-broadcast

* fix elementwise broadcast

* add broadcast for gpu elementsie

* feat: pad 支持 axes 负数

feat: 不支持的 padding 导出为独立的 pad 算子

feat: 支持导入 onnxsim 过的 inception
Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: 修正池化的测试

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 导出 pads，支持 inception 导入导出，已加入 ci

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 支持 densenet 导入导出，并加入 ci

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 导入 squeeze

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix softmax

* feat: 导出 clip 和 transpose

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 支持 Conv 的 bias

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: bias of conv

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: bias of conv

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 导入 split

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 导出 split

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: conv

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: conv group

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: matmul 的 bias 没有放在输入里，修正

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix exmaple

* fix: 改正 reduce_mean 导出

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* refactor: 修改 slice 实现与 onnx 一致

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* style: 不导出两个 runtime 函数

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* doc: 中文使用指南

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* doc: 补全指南

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: 修复导入数据的问题

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fmt

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 添加 Dropout 基本结构，但不支持两个输出是不同的类型

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 重新导出优化接口

feat: dropout 导入
Signed-off-by: YdrMaster <ydrml@hotmail.com>

* build: BANG 选项加入 Makefile

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fxi code, change of test/kernels/bang/test* is use NativeCpuRuntime.
chaneg of include/bang/bang_runtime is for the cntoolkit upgrade.

* feat: 导出 bang runtime

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* add USE_BANG=1

* fix matmul

* fix reshape

* fix

* fix activation

* fix transpose

* format

* format

* update Makefile

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 支持导入导出 ConvTranspose

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* add prelu on mlu

* fix: ConvTranspose

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* feat: 支持导入导出 PRelu

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* add convtrans on mlu

* fmt

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* docs: 更新 README_CN.md

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix code by review suggestions

* style

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix: Softmax 的 axis 可以用默认值？感觉是 onnx 不标准

Signed-off-by: YdrMaster <ydrml@hotmail.com>

* fix cuda & intelcpu bugs after merging

---------

Signed-off-by: YdrMaster <ydrml@hotmail.com>
Co-authored-by: wanghailu <wanghailu0717@163.com>
Co-authored-by: wanghailu <wanghailu@qiyuanlab.com>
Co-authored-by: whjthu <haojie0429@gmail.com>
 2023-04-18 15:10:33 +08:00
zhengly123 a1974aabcd
NNET supports TVM backend and kernels (#78) 
* Add: mutator InfoGAN minimum test

* Add: cache and padding (bugs!!)

* Add: expression reader as a cmake target

* Fix: [Intermediate] NMutator::expressionToGraph

To be fix: matmul with implicit broadcast

* Add: matmul broadcast

* Fix: GraphObj ctor should use cloneTensor

* Fix: cuBLAS failure when codegen is enabled

* Add: Exception for checkCuError

* Fix: graph OpList ctor

* Add: expr simplication for TVM

* Add: TVM headers and CMake include paths

* Add: CMake config

* Add: PackedFunc (broken)

* Fix: remove cuCtxCreate which makes TVM fails

* Fix: membound_tvm

* Fix: test_memboundOp

* Add: PRelu Expr and AsTVMVisitor

* Add: Random generator

* Add: support TVM packed function

* Fix: specify runtime

* Add: CMake support of TVM

* Add: detailed output of Matmul

* Add: comments for Matmul

* Chore: format and comments

* Chore: GraphObj::selfCheck without assert control

* Fix: CMAKE_CXX_FLAGS in CMakeLists

* fix merge bug

* update api for mkl batchnorm test

* fix lotus env

* fig header bug

---------

Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
Co-authored-by: huangshuhong <huangsh19@mails.tsinghua.edu.cn>
Co-authored-by: whjthu <haojie0429@gmail.com>
 2023-04-18 00:26:36 +08:00
wendy12022 43d4798323
ADD: sub graph replacement. (#56) 
reconfig: connections among op and tensor now is managered by GraphObj .

add some comments

merge from master

merge from master

ADD: sub graph replacement

reconfig inputs of op resize, due to the check of operator inputs.

ResizeObj::clone

clang format

fix some and add test for multi-output.

replacement support multi-inputs and multi-outputs.

add clone for all operators

add replaceSubGraph addSubGraph

remove extra code

add more test

remove extra print

Co-authored-by: Haojie Wang <haojie0429@gmail.com>
 2023-04-17 13:09:07 +08:00
wendy12022 c8b2c8ed32
Cpu backend2 (#77) 
fix review

change Device::MKL to Device::INTELCPU

fix mkl linkage

fix errors according to merge from master

now can call mkl backend

fix softmax/flatten with axis from onnx.

modify README.md

fix memory refree

add env_lotus_intelcpu.sh

fix compile

merge from branch cpu_backend

fix something add gather

fix something

FIX: directory rename from "mkl" to "intelcpu"

ADD: use oneMKL dpcpp interface to implement matmul kernel.

ADD: add dpcpp as compiler for mkl, and fix warnings for clang compiling.
add dpcpp kernel for pow.

ADD: mkl kernel for pad.

ADD: slice mkl kernel.

ADD: reshape/flatten/identity mkl kernel.

ADD: split mkl kernel.

fix compile error

FIX: fix flattenObj with axis.

ADD reduce_mean mkl kernel.

Add concat mkl kernel.

bathNorm for mkl kernel.

sigmoid mkl kernel.

ADD：add mkl kernel for pooling

add more tests for softmax

Now softmax cuda kernel supports any axises.

mkl kernel for softmax

softmax

add axis to softmax operator

add mkl kernel for abs tanh

ADD: relu kernel for mkl

fix binary mkl primitives.

add mkl kernel for binary operators

fix compiler error

move stream to runtime

clang format

add MemoryFormat for tensorObj.

use post_ops for fused conv/deconv

Distinguish mkl  op_timer from cuda op timer.

add act optype to conv and deconv

add operator timer

add mkl kernel for convTransposed

minor fix for group conv

do not use cblas_sgemm_batch

CpuRuntimeObj->NativeCpuRuntimeObj

add  matmul op for mkl
 2023-04-17 12:15:23 +08:00
Hardy fe1afe38fa
fix code of bang conv (#76) 
* fix code of bang conv

* test: 向 master push 时也执行 ci

Signed-off-by: YdrMaster <ydrml@hotmail.com>

---------

Signed-off-by: YdrMaster <ydrml@hotmail.com>
Co-authored-by: wanghailu <wanghailu@qiyuanlab.com>
Co-authored-by: YdrMaster <ydrml@hotmail.com>
 2023-03-29 15:47:32 +08:00
Hardy 823e66a9ff
Support perf bang 1115 (#57) 
* support matmul

* add matmul

* add matmul

* add code for cnnl matmul operation and test

* add conv

* add code for conv test on mlu

* add code for test cnnl conv on mlu

* add code for perf conv and matmul on mlu

* clang format

* fix convolution operation

* fxi cmaklist

* code format

* fix code

* code format

---------

Co-authored-by: wanghailu <wanghailu@qiyuanlab.com>
Co-authored-by: wanghailu <wanghailu0717@163.com>
 2023-03-29 13:52:56 +08:00
wendy12022 86ec4036ce
ADD: add mkl runtime for intel cpu , and add mkl kernel for matmul/conv/convtransposed. (#61) 
* move memory format transformation to TensorObj

clang format

add MemoryFormat for tensorObj.

use post_ops for fused conv/deconv

Distinguish mkl  op_timer from cuda op timer.

add act optype to conv and deconv

add operator timer

add mkl kernel for convTransposed

minor fix for group conv

do not use cblas_sgemm_batch

CpuRuntimeObj->NativeCpuRuntimeObj

add  matmul op for mkl

* fix: fix bugs when rebasing from master

fix: fix bugs when rebasing from master

* fix: update api after rebasing

* fix: fix format; fix onnx import

* fix: fix clang-format

* [fix] fix conv_transpose test

* [fix] use stronger test case for transposed conv

* [fix] remove tensor memory format; fix mkl transpose conv

* [fix] add FIXME tag for op_timer python api

---------

Co-authored-by: whjthu <haojie0429@gmail.com>
 2023-03-27 21:28:49 +08:00
whjthu d9886e9de3 fix: remove inline keyword in class; rename getter and setter for inputOf and outputOf 2023-03-25 12:04:24 +08:00
YdrMaster 5aeacedab3 fix: 从模板导出每个类型的 python 接口 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-22 09:46:40 +08:00
YdrMaster 9db97eb212 refactor: 整合操作张量数据的方法 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-21 14:00:04 +08:00
YdrMaster e294e46436 feat: 导出 pool 到 onnx 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster 40fb8390b1 feat: 导入时保存权重 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster a5e692baea feat: 导出 batchnorm 到 onnx 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster 5b6698bac7 feat: 导出全图的输出张量到 onnx 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster 3d122aebfe feat: 支持导出浮点向量 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster f44a4daf70 feat: 导出未初始化的张量 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster ed81861375 temp: 实现初始值导入，但 resnet 报错 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 17:23:32 +08:00
YdrMaster 71a87c27d1 feat: 导出 ReduceMean 到 onnx 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 15:09:12 +08:00
YdrMaster 2a23669394 feat: 导出 Reshape 到 onnx 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 15:09:12 +08:00
YdrMaster fe81fccf76 feat: 导出 OperatorObj 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 15:09:12 +08:00
YdrMaster 45a3cdfa30 feat: GraphObj 增加一个拓扑排序方法及其测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 15:09:12 +08:00
YdrMaster f20e791cf5 style: 修改 graph.h/graph.cc 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-03-15 15:09:12 +08:00
Haojie Wang 0f52d04882
Merge branch 'master' into dev-onnx 2023-03-15 14:52:03 +08:00
deathwings602 40d1b1c91b
Add ConvTransposedNHWC (#67) 
* Add: IT_ASSERT_TODO

* [WIP] Add: ConvTranspose2d mutation test

* add ConvTransposedNHWC

* fix test_cuda_transposed_2d

---------

Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
Co-authored-by: huangshuhong <huangsh19@mails.tsinghua.edu.cn>
 2023-03-01 14:15:02 +08:00
YdrMaster 6871fff02b feat: 导出分配内存和运行推理的接口 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-23 11:08:00 +08:00
YdrMaster 4c7fdf44c5 feat: 前端支持 Conv 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-22 15:05:44 +08:00
YdrMaster 6a4de807e6 style: remove non-ascii comments from cpp 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-16 14:57:51 +08:00
YdrMaster 315763a83a feat: 前端支持 pad 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-15 11:41:06 +08:00
YdrMaster 8fae67b4b4 feat: 前端支持 slice 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 17:35:18 +08:00
YdrMaster f9d0076a86 opt: 优化 SliceObj 构造器实现 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 16:44:08 +08:00
YdrMaster 341cf1f943 feat: 前端支持 pool 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 16:26:47 +08:00
YdrMaster 62ceb78ae3 feat: 前端支持 reduceMean 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 15:35:01 +08:00
YdrMaster fb9d84dbb7 opt: 优化 ReduceMeanObj 构造器实现 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 15:14:28 +08:00
YdrMaster d11fb0ad5f feat: 前端支持 gather 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 14:16:01 +08:00
YdrMaster 45aa0237da feat: 前端支持 concat 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 13:42:35 +08:00
YdrMaster a7e58bd8d0 feat: 补充 DataType 类型 
- 增加了 6 个代数类型，与 onnx 的序号对应
- 现在可以导入 reshape 了

Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 11:27:57 +08:00
YdrMaster 7626efbfa8 feat: 前端支持 reshape 
- 无法测试，因为后端不支持 shape 的 INT64 类型

opt: ReshapeObj 构造改为全部传值并在内部 move
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-14 09:51:11 +08:00
whjthu 26be533faa Add documentation for operators. 2023-02-13 22:51:15 +08:00
YdrMaster cca4d2a491 feat: 前端支持 batchNorm（无单元测试） 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-13 17:15:35 +08:00
YdrMaster e194dd943b feat: 前端支持 flatten 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-13 13:50:07 +08:00
YdrMaster e4ec9c4230 feat: 前端支持 identity 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-13 12:26:11 +08:00
YdrMaster 7f0c8ebae3 feat: 前端支持 relu sigmoid tanh softmax abs 及单元测试 
Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-13 11:54:54 +08:00
YdrMaster 6e5beceadd feat: 增加 add sub mul div pow 前端 
- 添加每个算子的单元测试
- 添加线性回归模型导入测试

Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-13 11:25:54 +08:00
YdrMaster 296fcc5aa0 feat: 创建 pyinfinitensor 前端 
- python 前端项目结构及打包和安装脚本
- 后端编译出 so 改名为 backend，增加 GraphHandler 修改图结构
- ci 支持测试这些功能

Signed-off-by: YdrMaster <ydrml@hotmail.com>
 2023-02-13 09:19:05 +08:00
zhengly123 c7ec9ee6e7
Add search engine (#64) 
* Add: tensor fuid

* [Intermediate state] Add: Graph ctor for OpVec

* Add: clone for operators

* tmp: search_engine

* search: init search Engine.

* Add: dummy mutator for the test of search engine

* search: add print graph.

* search: add partition.

* search: update comments.

* Fix: remain FUID in Tensor::clone

* Chore: rename GUidBaseType to UidBaseType

* Fix: connect NMutator to SearchEngine

* Chore: output

* Fix test_memboundOp: nmutator uses input runtime

* Chore: clang-format

* Chore: clang-format

* Fix: comments in the review

---------

Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
Co-authored-by: mazx <dyxdy@live.com>
 2023-02-12 18:27:52 +08:00
wendy12022 d780f687fc
ADD: reconfig ResizeObj, support "tf_crop_and_resize " and cubic coeff kernel. (#59) 
add cubic coef

add tf_crop_and_resize
 2022-12-24 04:02:21 +08:00
wendy12022 c5966f8d81
Add: resize operator and cuda kernel,support nearest/linear coef. (#51) 
ADD: resize operator and cuda kernel,support nearest/linear coef.

fix some

fix tests

add more tests for linear mode.

add linear coef mode.

add scales

add tests

fix tests.

add notLarger notSmaller

fix

add test

ADD:resize operator and cuda kernel
 2022-11-14 09:30:22 +08:00
zhengly123 63d8aff985
Fix: cuCtxCreate before other initialization (#49) 
Fix: create cuCtx at the very beginning

Co-authored-by: Liyan Zheng <liyan-zheng@outlook.com>
 2022-10-19 15:41:48 +08:00