openvino/docs/articles_en/learn-openvino/openvino-samples/throughput-benchmark.rst

179 lines
5.5 KiB
ReStructuredText

.. {#openvino_sample_throughput_benchmark}
Throughput Benchmark Sample
===========================
.. meta::
:description: Learn how to estimate performance of a model using Asynchronous Inference Request API in throughput mode (Python, C++).
This sample demonstrates how to estimate performance of a model using Asynchronous
Inference Request API in throughput mode. Unlike `demos <https://docs.openvino.ai/2024/omz_demos.html>`__ this sample
does not have other configurable command-line arguments. Feel free to modify sample's
source code to try out different options.
The reported results may deviate from what :doc:`benchmark_app <benchmark-tool>`
reports. One example is model input precision for computer vision tasks. benchmark_app
sets ``uint8``, while the sample uses default model precision which is usually ``float32``.
Before using the sample, refer to the following requirements:
- The sample accepts any file format supported by ``core.read_model``.
- The sample has been validated with: `yolo-v3-tf <https://docs.openvino.ai/2024/omz_models_model_yolo_v3_tf.html>`__,
`face-detection-0200 <https://docs.openvino.ai/2024/omz_models_model_face_detection_0200.html>`__ models.
- To build the sample, use instructions available at :ref:`Build the Sample Applications <build-samples>`
section in "Get Started with Samples" guide.
How It Works
####################
The sample compiles a model for a given device, randomly generates input data,
performs asynchronous inference multiple times for a given number of seconds.
Then, it processes and reports performance results.
.. tab-set::
.. tab-item:: Python
:sync: python
.. scrollbox::
.. doxygensnippet:: samples/python/benchmark/throughput_benchmark/throughput_benchmark.py
:language: python
.. tab-item:: C++
:sync: cpp
.. scrollbox::
.. doxygensnippet:: samples/cpp/benchmark/throughput_benchmark/main.cpp
:language: cpp
You can see the explicit description of each sample step at
:doc:`Integration Steps <../../openvino-workflow/running-inference/integrate-openvino-with-your-application>`
section of "Integrate OpenVINO™ Runtime with Your Application" guide.
Running
####################
.. tab-set::
.. tab-item:: Python
:sync: python
.. code-block:: console
python throughput_benchmark.py <path_to_model> <device_name>(default: CPU)
.. tab-item:: C++
:sync: cpp
.. code-block:: console
throughput_benchmark <path_to_model> <device_name>(default: CPU)
To run the sample, you need to specify a model. You can get a model specific for
your inference task from one of model repositories, such as TensorFlow Zoo, HuggingFace, or TensorFlow Hub.
Example
++++++++++++++++++++
1. Download a pre-trained model.
2. You can convert it by using:
.. tab-set::
.. tab-item:: Python
:sync: python
.. code-block:: python
import openvino as ov
ov_model = ov.convert_model('./models/googlenet-v1')
# or, when model is a Python model object
ov_model = ov.convert_model(googlenet-v1)
.. tab-item:: CLI
:sync: cli
.. code-block:: console
ovc ./models/googlenet-v1
3. Perform benchmarking, using the ``googlenet-v1`` model on a ``CPU``:
.. tab-set::
.. tab-item:: Python
:sync: python
.. code-block:: console
python throughput_benchmark.py ./models/googlenet-v1.xml
.. tab-item:: C++
:sync: cpp
.. code-block:: console
throughput_benchmark ./models/googlenet-v1.xml
Sample Output
####################
.. tab-set::
.. tab-item:: Python
:sync: python
The application outputs performance results.
.. code-block:: console
[ INFO ] OpenVINO:
[ INFO ] Build ................................. <version>
[ INFO ] Count: 2817 iterations
[ INFO ] Duration: 10012.65 ms
[ INFO ] Latency:
[ INFO ] Median: 13.80 ms
[ INFO ] Average: 14.10 ms
[ INFO ] Min: 8.35 ms
[ INFO ] Max: 28.38 ms
[ INFO ] Throughput: 281.34 FPS
.. tab-item:: C++
:sync: cpp
The application outputs performance results.
.. code-block:: console
[ INFO ] OpenVINO:
[ INFO ] Build ................................. <version>
[ INFO ] Count: 1577 iterations
[ INFO ] Duration: 15024.2 ms
[ INFO ] Latency:
[ INFO ] Median: 38.02 ms
[ INFO ] Average: 38.08 ms
[ INFO ] Min: 25.23 ms
[ INFO ] Max: 49.16 ms
[ INFO ] Throughput: 104.96 FPS
Additional Resources
####################
- :doc:`Integrate the OpenVINO™ Runtime with Your Application <../../openvino-workflow/running-inference/integrate-openvino-with-your-application>`
- :doc:`Get Started with Samples <get-started-demos>`
- :doc:`Using OpenVINO Samples <../openvino-samples>`
- :doc:`Convert a Model <../../documentation/legacy-features/transition-legacy-conversion-api/legacy-conversion-api>`
- `Throughput Benchmark Python Sample on Github <https://github.com/openvinotoolkit/openvino/blob/master/samples/python/benchmark/throughput_benchmark/README.md>`__
- `Throughput Benchmark C++ Sample on Github <https://github.com/openvinotoolkit/openvino/blob/master/samples/cpp/benchmark/throughput_benchmark/README.md>`__