Note: This is an beta (preview) version which is still under refining.

nn-Meter is a novel and efficient system to accurately predict the inference latency of DNN models on diverse edge devices. The key idea is dividing a whole model inference into kernels, i.e., the execution units of fused operators on a device, and conduct kernel-level prediction. We currently evaluate four popular platforms on a large dataset of 26k models. It achieves 99.0% (mobile CPU), 99.1% (mobile Adreno 640 GPU), 99.0% (mobile Adreno 630 GPU), and 83.4% (Intel VPU) prediction accuracy.

The current supported hardware and inference frameworks:

Device	Framework	Processor	+-10% Accuracy	Hardware name
Pixel4	TFLite v2.1	CortexA76 CPU	99.0%	cortexA76cpu_tflite21
Mi9	TFLite v2.1	Adreno 640 GPU	99.1%	adreno640gpu_tflite21
Pixel3XL	TFLite v2.1	Adreno 630 GPU	99.0%	adreno630gpu_tflite21
Intel Movidius NCS2	OpenVINO2019R2	Myriad VPU	83.4%	myriadvpu_openvino2019r2

nn-Meter has achieved the Mobisys 21 Best Paper Award! For more details, please check out paper:

nn-Meter: towards accurate latency prediction of deep-learning model inference on diverse edge devices

Who should consider using nn-Meter

• Those who want to get the DNN inference latency on mobile and edge devices with no deployment efforts on real devices.
• Those who want to run hardware-aware NAS with NNI.
• Those who want to build latency predictors for their own devices.
• Those who want to use the 26k latency benchmark dataset.

Installation

Currently nn-Meter has been tested on Linux and Windows system. Windows 10, Ubuntu 16.04 and 20.04 with python 3.6.10 are tested and supported. Please first install python3 before nn-Meter installation. Then nn-Meter could be installed by running:

pip install nn-meter

If you want to try latest code, please install nn-Meter from source code. First git clone nn-Meter package to local:

git clone git@github.com:microsoft/nn-Meter.git
cd nn-Meter

Then simply run the following pip install in an environment that has python >= 3.6. The command will complete the automatic installation of all necessary dependencies and nn-Meter.

pip install .

nn-Meter is a latency predictor of models with type of Tensorflow, PyTorch, Onnx, nn-meter IR graph and NNI IR graph. To use nn-Meter for specific model type, you also need to install corresponding required packages. The well tested versions are listed below:

Testing Model Type	Requirements
Tensorflow	tensorflow==2.6.0
Torch	torch==1.9.0, torchvision==0.10.0, (alternative)[onnx==1.9.0, onnx-simplifier==0.3.6] or [nni>=2.4][1]
Onnx	onnx==1.9.0
nn-Meter IR graph	---
NNI IR graph	nni>=2.4

[1] Please refer to nn-Meter Usage for more information.

Please also check the versions of numpy and scikit_learn. The different versions may change the prediction accuracy of kernel predictors.

The stable version of wheel binary package will be released soon.

Usage

To apply for hardware latency prediction, nn-Meter provides two types of interfaces：

• command line nn-meter after nn-meter installation.
• Python binding provided by the module nn_meter

Here is a summary of supported inputs of the two methods.

Testing Model Type	Command Support	Python Binding
Tensorflow	Checkpoint file dumped by tf.saved_model() and end with .pb	Checkpoint file dumped by tf.saved_model and end with .pb
Torch	Models in torchvision.models	Object of torch.nn.Module
Onnx	Checkpoint file dumped by torch.onnx.export() or onnx.save() and end with .onnx	Checkpoint file dumped by onnx.save() or model loaded by onnx.load()
nn-Meter IR graph	Json file in the format of nn-Meter IR Graph	dict object following the format of nn-Meter IR Graph
NNI IR graph	-	NNI IR graph object

In both methods, users could appoint predictor name and version to target a specific hardware platform (device). Currently, nn-Meter supports prediction on the following four configs:

Predictor (device_inferenceframework)	Processor Category	Version
cortexA76cpu_tflite21	CPU	1.0
adreno640gpu_tflite21	GPU	1.0
adreno630gpu_tflite21	GPU	1.0
myriadvpu_openvino2019r2	VPU	1.0

Users can get all predefined predictors and versions by running

# to list all predefined predictors
nn-meter --list-predictors 

Predict latency of saved CNN model

After installation, a command named nn-meter is enabled. To predict the latency for a CNN model with a predefined predictor in command line, users can run the following commands

# for Tensorflow (*.pb) file
nn-meter predict --predictor <hardware> [--predictor-version <version>] --tensorflow <pb-file_or_folder> 

# for ONNX (*.onnx) file
nn-meter predict --predictor <hardware> [--predictor-version <version>] --onnx <onnx-file_or_folder>

# for torch model from torchvision model zoo (str)
nn-meter predict --predictor <hardware> [--predictor-version <version>] --torchvision <model-name> <model-name>... 

# for nn-Meter IR (*.json) file
nn-meter predict --predictor <hardware> [--predictor-version <version>] --nn-meter-ir <json-file_or_folder> 

--predictor-version <version> arguments is optional. When the predictor version is not specified by users, nn-meter will use the latest version of the predictor.

nn-Meter can support batch mode prediction. To predict latency for multiple models in the same model type once, user should collect all models in one folder and state the folder after --[model-type] liked argument.

It should also be noted that for PyTorch model, nn-meter can only support existing models in torchvision model zoo. The string followed by --torchvision should be exactly one or more string indicating name(s) of some existing torchvision models. To apply latency prediction for torchvision model in command line, onnx and onnx-simplifier packages are required.

Convert to nn-Meter IR Graph

Furthermore, users may be interested to convert tensorflow pb-file or onnx file to nn-Meter IR graph. Users could convert nn-Meter IR graph and save to .json file be running

# for Tensorflow (*.pb) file
nn-meter get_ir --tensorflow <pb-file> [--output <output-name>]

# for ONNX (*.onnx) file
nn-meter get_ir --onnx <onnx-file> [--output <output-name>]

Output name is default to be /path/to/input/file/<input_file_name>_<model-type>_ir.json if not specified by users.

Use nn-Meter in your python code

After installation, users can import nn-Meter in python code

from nn_meter import load_latency_predictor

predictor = load_latency_predictor(hardware_name, hardware_predictor_version) # case insensitive in backend

# build your model (e.g., model instance of torch.nn.Module)
model = ... 

lat = predictor.predict(model, model_type) # the resulting latency is in unit of ms

By calling load_latency_predictor, user selects the target hardware and loads the corresponding predictor. nn-Meter will try to find the right predictor file in ~/.nn_meter/data. If the predictor file doesn't exist, it will download from the Github release.

In predictor.predict(), the allowed items of the parameter model_type include ["pb", "torch", "onnx", "nnmeter-ir", "nni-ir"], representing model types of tensorflow, torch, onnx, nn-meter IR graph and NNI IR graph, respectively.

<span id="torch-model-converters"> For Torch models, the shape of feature maps is unknown merely based on the given network structure, which is, however, significant parameters in latency prediction. Therefore, torch model requires a shape of input tensor for inference as a input of predictor.predict(). Based on the given input shape, a random tensor according to the shape will be generated and used. Another thing for Torch model prediction is that users can install the onnx and onnx-simplifier packages for latency prediction (referred to as Onnx-based latency prediction for torch model), or alternatively install the nni package (referred to as NNI-based latency prediction for torch model). Note that the nni option does not support command line calls. In addition, if users use nni for latency prediction, the PyTorch modules should be defined by the nn interface from NNI import nni.retiarii.nn.pytorch as nn (view NNI doc for more information), and the parameter apply_nni should be set as True in the function predictor.predict(). Here is an example of NNI-based latency prediction for Torch model:

import nni.retiarii.nn.pytorch as nn
from nn_meter import load_latency_predictor

predictor = load_latency_predictor(...)

# build your model using nni.retiarii.nn.pytorch as nn
model = nn.Module ...

input_shape = (1, 3, 224, 224)
lat = predictor.predict(model, model_type='torch', input_shape=input_shape, apply_nni=True) 

The Onnx-based latency prediction for torch model is stable but slower, while the NNI-based latency prediction for torch model is unstable as it could fail in some case but much faster compared to the Onnx-based model. The Onnx-based model is set as the default one for Torch model latency prediction in nn-Meter. Users could choose which one they preferred to use according to their needs.

Users could view the information all built-in predictors by list_latency_predictors or view the config file in nn_meter/configs/predictors.yaml.

Users could get a nn-Meter IR graph by applying model_file_to_graph and model_to_graph by calling the model name or model object and specify the model type. The supporting model types of model_file_to_graph include "onnx", "pb", "torch", "nnmeter-ir" and "nni-ir", while the supporting model types of model_to_graph include "onnx", "torch" and "nni-ir".

Benchmark Dataset

To evaluate the effectiveness of a prediction model on an arbitrary DNN model, we need a representative dataset that covers a large prediction scope. As there is no such available latency dataset, nn-Meter collects and generates 26k CNN models. It contains various operators, configurations, and edge connections, with covering different levels of FLOPs and latency. (Please refer the paper for the dataset generation method and dataset numbers.)

We release the dataset, and provide an interface of nn_meter.dataset for users to get access to the dataset. Users can also download the data from the Download Link for testing nn-Meter or their own prediction models.

Hardware-aware NAS by nn-Meter and NNI

To empower affordable DNN on the edge and mobile devices, hardware-aware NAS searches both high accuracy and low latency models. In particular, the search algorithm only considers the models within the target latency constraints during the search process.

Currently we provide two examples of hardware-aware NAS, including end-to-end multi-trial NAS which is a random search algorithm on SPOS NAS search space, and the popular ProxylessNAS, which is a one-shot NAS algorithm with hardware-efficient loss function. More examples of other widely-used hardware-aware NAS and model compression algorithms are coming soon.

Multi-trial SPOS Demo

To run multi-trail SPOS demo, NNI should be installed through source code by following NNI Doc

python setup.py develop

Then run multi-trail SPOS demo:

python ${NNI_ROOT}/examples/nas/oneshot/spos/multi_trial.py

How the demo works

Refer to NNI Doc for how to perform NAS by NNI.

To support hardware-aware NAS, you first need a Strategy that supports filtering the models by latency. We provide such a filter named LatencyFilter in NNI and initialize a Random strategy with the filter:

simple_strategy = strategy.Random(model_filter=LatencyFilter(threshold=100, predictor=base_predictor))

LatencyFilter will predict the models' latency by using nn-Meter and filter out the models whose latency with the given predictor are larger than the threshold (i.e., 100 in this example). You can also build your own strategies and filters to support more flexible NAS such as sorting the models according to latency.

Then, pass this strategy to RetiariiExperiment:

exp = RetiariiExperiment(base_model, trainer, strategy=simple_strategy)

exp_config = RetiariiExeConfig('local')
...
exp_config.dummy_input = [1, 3, 32, 32]

exp.run(exp_config, port)

In exp_config, dummy_input is required for tracing shape info.

ProxylessNAS Demo

To run the one-shot ProxylessNAS demo, users can run the NNI ProxylessNAS training demo:

python ${NNI_ROOT}/examples/nas/oneshot/proxylessnas/main.py --applied_hardware <hardware> --reference_latency <reference latency (ms)>

How the demo works

Refer to NNI Doc for how to perform NAS by NNI.

ProxylessNAS currently builds a lookup table, that stores the measured latency of each candidate building block in the search space. The latency sum of all building blocks in a candidate model will be treated as the model inference latency. With leveraging nn-Meter in NNI, users can apply ProxylessNAS to search efficient DNN models on more types of edge devices. In NNI implementation, a HardwareLatencyEstimator predicts expected latency for the mixed operation based on the path weight of ProxylessLayerChoice. To call nn-Meter in NNI ProxylessNAS, users can add the arguments of "--applied_hardware <hardware> --reference_latency <reference latency (ms)>" in the example.

Bench Dataset

To evaluate the effectiveness of a prediction model on an arbitrary DNN model, we need a representative dataset that covers a large prediction scope. nn-Meter collects and generates 26k CNN models. (Please refer the paper for the dataset generation method.)

We release the dataset, and provide an interface of nn_meter.dataset for users to get access to the dataset. Users can also download the data from the Download Link on their own.

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

License

The entire codebase is under MIT license

The dataset is under Open Use of Data Agreement

Citation

If you find that nn-Meter helps your research, please consider citing it:

@inproceedings{nnmeter,
    author = {Zhang, Li Lyna and Han, Shihao and Wei, Jianyu and Zheng, Ningxin and Cao, Ting and Yang, Yuqing and Liu, Yunxin},
    title = {nn-Meter: Towards Accurate Latency Prediction of Deep-Learning Model Inference on Diverse Edge Devices},
    year = {2021},
    publisher = {ACM},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3458864.3467882},
    doi = {10.1145/3458864.3467882},
    booktitle = {Proceedings of the 19th Annual International Conference on Mobile Systems, Applications, and Services},
    pages = {81–93},
}

@misc{nnmetercode,
    author = {Microsoft Research nn-Meter Team},
    title = {nn-Meter: Towards Accurate Latency Prediction of Deep-Learning Model Inference on Diverse Edge Devices},
    year = {2021},
    url = {https://github.com/microsoft/nn-Meter},
}