Speed up inference of forecaster through OpenVINO

Introduction

In the inferencing process, it is desirable to speed up. One way to do this is to utilize some accelerators, such as OpenVINO.

Actually, utilizing OpenVINO to accelerate is easy in Chronos, that is directly calling predict_with_openvino (optionally build_openvino). In this guidance, we demonstrate how to speed up inference of forecaster through OpenVINO in detail.

We will take TCNForecaster and nyc_taxi dataset as an example in this guide.

Setup

Before we begin, we need to install chronos if it isn’t already available, we choose to use pytorch as deep learning backend.

!pip install --pre --upgrade bigdl-chronos[pytorch]
# install OpenVINO
!pip install openvino-dev
# fix conflict with google colab
!pip uninstall -y torchtext
!pip install numpy==1.21

📝Note

• Although Chronos supports inferencing on a cluster, the method to speed up can only be used when forecaster is a non-distributed version.

• Only pytorch backend deep learning forecasters support openvino acceleration.

Forecaster preparation

Before the inferencing process, a forecaster should be created and trained. The training process is introduced in the previous guidance Train forcaster on single node in detail, therefore we directly create and train a TCNForecaster based on the nyc taxi dataset.

Speeding up inference

When a trained forecaster is ready and forecaster is a non-distributed version, we provide with predict_with_openvino method to speed up inference. The method can be directly called without calling build_openvino and forecaster will automatically build an openvino session with default settings.

📝Note

build_openvino is recommended to use in following cases:

1. To strictly control the thread to be used during inferencing.

2. To alleviate the cold start problem when predict_with_openvino is called for the first time.

Please refer to API documentation for more information on build_openvino.

The predict_with_openvino method currently only supports data in numpy ndarray format and the parameter batch_size can be changed. If not familiar with manual hyperparameters tuning, just leave batch_size to the default value.

# get data for training and testing
train_data, test_data = get_data()
# get a trained forecaster
forecaster = get_trained_forecaster(train_data)

# speed up inference through OpenVINO
for x, y in test_data:
    yhat = forecaster.predict_with_openvino(x.numpy()) # predict

Let’s see the acceleration performance of predict_with_openvino.

The predict latency of without accelerator and with OpenVINO are given below. The result “p50” means latency sorted to 50% in multiple predictions and the acceleration performance is significant.

from bigdl.chronos.metric import Evaluator

x = next(iter(test_data))[0]
def func_original():
    forecaster.predict(x.numpy()) # without accelerator
def func_openvino():
    forecaster.predict_with_openvino(x.numpy()) # with OpenVINO

print("original predict runtime (ms):", Evaluator.get_latency(func_original))
print("pridict runtime with OpenVINO (ms):", Evaluator.get_latency(func_openvino))