How to Use AI Benchmark¶

The tutorial will guide you how to use AI Benchmark on ROScube series.

Introduction¶

We used AI Benchmark to evaluate GPU performance on ROScube series.

AI Benchmark Alpha is an open source python library for evaluating AI performance of various hardware platforms, including CPUs, GPUs and TPUs.

Here, we provided two usages on different ROScube series.

Note

GPU with at least 2GB of RAM is required for running inference tests / 4GB of RAM for training tests.

The benchmark is compatible with both TensorFlow 1.x and 2.x versions.

pip install ai-benchmark

from ai_benchmark import AIBenchmark
results = AIBenchmark().run()

To run inference or training only, use benchmark.run_inference() or benchmark.run_training().

Requirement:

Python: 3.8

Keras: 2.6

TensorFlow: 2.6

Cuda: 11.4

CudNN: 8.2

Nvidia-driver: >= 470

pip install tensorflow==<version>

pip install ai-benchmark

from ai_benchmark import AIBenchmark
results = AIBenchmark().run()

To run inference or training only, use benchmark.run_inference() or benchmark.run_training().

In total, AI Benchmark consists of 42 tests and 19 sections.

After testing, you will get the Score of GPU performance:

Inference Score

Training Score

AI Score

Then go to the ranking page, and you can compare your device with open data.

And, we provided some testing data of ROScube:

When you run the python code, but you can’t show the CUDA version, like N/A.

gedit ~/.bashrc

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

source ~/.bashrc
nvcc -V