oaid / mxnet-hrt Goto Github PK

Heterogeneous Run Time version of MXNet. Added heterogeneous capabilities to the MXNet, uses heterogeneous computing infrastructure framework to speed up Deep Learning on Arm-based heterogeneous embedded platform. It also retains all the features of the original MXNet architecture which users deploy their applications seamlessly.

License: Apache License 2.0

CMake 0.49% Makefile 0.43% R 2.54% C++ 32.65% Python 28.11% Java 0.03% Shell 1.20% C 0.88% Jupyter Notebook 13.02% Cuda 5.19% Batchfile 0.11% MATLAB 0.32% Perl 6.28% Scala 8.71% Perl 6 0.04% Rebol 0.01%

arm arm-compute-library arm-neon arm-gpu machine-learning artificial-intelligence mxnet opencl dnn cnn

mxnet-hrt's Introduction

MXNet-HRT

MXNet-HRT is a project that is maintained by OPEN AI LAB, it uses Arm Compute Library (NEON+GPU) to speed up MXNet and provide utilities to debug, profile and tune application performance.

The release version is 0.3.1, is based on Rockchip RK3399 Platform, target OS is Ubuntu 16.04. Can download the source code from OAID/MXNet-HRT

The ARM Computer Vision and Machine Learning library is a set of functions optimised for both ARM CPUs and GPUs using SIMD technologies. See also Arm Compute Library.
MXNet is a Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more. See also MXNet.

Documents

Arm Compute Library Compatibility Issues :

There are some compatibility issues between ACL and MXNet Layers, we bypass it to MXNet's original layer class as the workaround solution for the below issues

Normalization in-channel issue
Tanh issue
Softmax supporting multi-dimension issue
Group issue

Performance need be fine turned in the future

Release History

The MXNet based version is 26b1cb9ad0bcde9206863a6f847455ff3ec3c266.

Version 0.3.1 - Feb 09, 2018

Support Arm Compute Library version 17.12

Version 0.3.0 - Jan 31, 2018

Support Arm Compute Library version 17.12

Version 0.2.0 - Aug 27, 2017

Support Arm Compute Library version 17.06 with 4 new layers added

Batch Normalization Layer
Direct convolution Layer
Concatenate layer

Version 0.1.0 - Jul 6, 2017

Initial version supports 10 Layers accelerated by Arm Compute Library version 17.05 :

Convolution Layer
Pooling Layer
LRN Layer
ReLU Layer
Sigmoid Layer
Softmax Layer
TanH Layer
AbsVal Layer
BNLL Layer
InnerProduct Layer

Issue Report

Encounter any issue, please report on issue report. Issue report should contain the following information :

The exact description of the steps that are needed to reproduce the issue
The exact description of what happens and what you think is wrong

mxnet-hrt's People

Contributors

Stargazers

Watchers

mxnet-hrt's Issues

Is it possible to disable OpenCL? only need NEON

For bugs or installation issues, please provide the following information.
The more information you provide, the more likely people will be able to help you.

Environment info

Operating System:

Compiler:

Package used (Python/R/Scala/Julia):

MXNet version:

Or if installed from source:

MXNet commit hash (git rev-parse HEAD):

If you are using python package, please provide

Python version and distribution:

If you are using R package, please provide

R sessionInfo():

Error Message:

Please paste the full error message, including stack trace.

Minimum reproducible example

if you are using your own code, please provide a short script that reproduces the error.

Steps to reproduce

or if you are running standard examples, please provide the commands you have run that lead to the error.

What have you tried to solve it?

How to make MXNet-HRT support higher version mxnet? such as mxnet 1.0 or 1.1?

when i was testin insightface project(a mxnet deep network),i got these errors:
firefly@firefly:~/zzp/insightface/deploy$ python test.py
loading ../models/model-r34-amf/model 0
[02:10:23] /home/firefly/MXNet-HRT/dmlc-core/include/dmlc/./logging.h:304: [02:1 0:23] /home/firefly/MXNet-HRT/nnvm/dmlc-core/include/dmlc/././json.h:838: JSONRe ader: Unknown field attrs, candidates are:
"attr"
"backward_source_id"
"control_deps"
"inputs"
"name"
"op"
"param"

**I guess it's because the mxnet version too low

how can i make the MXNet-HRT to support mxnet of hihger versions?**

MXNet-HRT vs Caffe.

Hi,
I am using FaceDetection with MXNet. It is running good with caffe but not with MXNet-HRT.
Procedure i did-

followed this link- https://github.com/OAID/MXNet-HRT/blob/master/acl_openailab/installation.md
Changed the frame from image to live feed from camera.
git status gives me -origin/master, but git status on submodules givre me different versions.

Any suggestion? Is there any specific versions for all needs to be there to work properly?
Thanks.

OpenCL GPU acceleration for mxnetOnACL

When trying mxnetOnACL on my ARM64 Linux box, I found that OpenCL was actually not used in the classification demo following the instructions at:

https://github.com/OAID/mxnetOnACL/tree/master/acl_openailab

Did I miss something?

Is there benchmark score for VGG or other models?

ssd

I follow the instruction of example/ssd and run python demo.py from example/ssd directory and got an error below. My platform is firefly 3399. Is there a reason which ssd example is not working?

[21:15:32] /home/firefly/2TB/src/firefly/mxnetOnACL/dmlc-core/include/dmlc/./logging.h:304: [21:15:32] src/core/symbolic.cc:72: Symbol.ComposeKeyword argument name anchors not found.
Candidate arguments:
[0]cls_prob
[1]loc_pred
[2]anchor

Stack trace returned 7 entries:
[bt] (0) /usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/libmxnet.so(_ZN4dmlc15LogMessageFatalD1Ev+0x44) [0x7f7a5b1cb4]
[bt] (1) /usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/libmxnet.so(_ZN4nnvm23KeywordArgumentMismatchEPKcRKSt6vectorINSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEESaIS8_EERKN4dmlc10array_viewIS8_EE+0x434) [0x7f7b095604]
[bt] (2) /usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/libmxnet.so(ZN4nnvm6Symbol7ComposeERKN4dmlc10array_viewIPKS0_EERKSt13unordered_mapINSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEES4_St4hashISE_ESt8equal_toISE_ESaISt4pairIKSE_S4_EEERSK+0xcc4) [0x7f7b0922cc]
[bt] (3) /usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/libmxnet.so(NNSymbolCompose+0x238) [0x7f7b0a1188]
[bt] (4) /usr/lib/aarch64-linux-gnu/libffi.so.6(ffi_call_SYSV+0x64) [0x7f7bfd1e60]
[bt] (5) /usr/lib/aarch64-linux-gnu/libffi.so.6(ffi_call+0xc0) [0x7f7bfd27b8]
[bt] (6) /usr/lib/python2.7/lib-dynload/_ctypes.aarch64-linux-gnu.so(_ctypes_callproc+0x670) [0x7f7bff1b30]

Traceback (most recent call last):
File "demo.py", line 99, in
ctx, args.nms_thresh, args.force_nms)
File "demo.py", line 40, in get_detector
.get_symbol(len(CLASSES), nms_thresh, force_nms)
File "/home/firefly/2TB/src/mxnetOnACL/example/ssd/symbol/symbol_vgg16_ssd_300.py", line 179, in get_symbol
net = get_symbol_train(num_classes)
File "/home/firefly/2TB/src/mxnetOnACL/example/ssd/symbol/symbol_vgg16_ssd_300.py", line 150, in get_symbol_train
variances=(0.1, 0.1, 0.2, 0.2), nms_topk=nms_topk)
File "", line 47, in _contrib_MultiBoxDetection
File "/usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/_ctypes/symbol.py", line 136, in _symbol_creator
s._compose(name=name, **kwargs)
File "/usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/symbol.py", line 419, in _compose
self.handle, name, num_args, keys, args))
File "/usr/local/lib/python2.7/dist-packages/mxnet-0.10.1-py2.7.egg/mxnet/base.py", line 85, in check_call
raise MXNetError(py_str(_LIB.MXGetLastError()))
mxnet.base.MXNetError: [21:15:32] src/core/symbolic.cc:72: Symbol.ComposeKeyword argument name anchors not found.
Candidate arguments:
[0]cls_prob
[1]loc_pred
[2]anchor

Thanks,

cannot make when build mxnet

I have followed the instruction.
However, it keeps failing during 3.2make mxnet "make" steps.
I have tried to compile computer library again but the library are still not compatible or missing.
Any help would be appreciated. Thanks

Environment info

Operating System: Ubtunu 16.04

Error Message:

/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 当搜索用于 /usr/lib/../lib/libopenblas.so 时跳过不兼容的 -lopenblas
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 当搜索用于 /usr/lib/../lib/libopenblas.a 时跳过不兼容的 -lopenblas
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 当搜索用于 //usr/lib/libopenblas.so 时跳过不兼容的 -lopenblas
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 当搜索用于 //usr/lib/libopenblas.a 时跳过不兼容的 -lopenblas
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 找不到 -lopenblas
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 找不到 -lOpenCL
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 找不到 -lopencv_core
/usr/lib/gcc-cross/aarch64-linux-gnu/5/../../../../aarch64-linux-gnu/bin/ld: 找不到 -lopencv_imgproc
collect2: error: ld returned 1 exit status
Makefile:256: recipe for target 'lib/libmxnet.so' failed
make: *** [lib/libmxnet.so] Error 1

oaid / mxnet-hrt Goto Github PK

mxnet-hrt's Introduction

MXNet-HRT

Documents

Arm Compute Library Compatibility Issues :

Release History

Version 0.3.1 - Feb 09, 2018

Version 0.3.0 - Jan 31, 2018

Version 0.2.0 - Aug 27, 2017

Version 0.1.0 - Jul 6, 2017

Issue Report

mxnet-hrt's People

Contributors

Stargazers

Watchers

Forkers

mxnet-hrt's Issues

Environment info

Error Message:

Minimum reproducible example

Steps to reproduce

What have you tried to solve it?

**I guess it's because the mxnet version too low

Environment info

Error Message:

Recommend Projects

Recommend Topics

Recommend Org