Synopsis

Benchmarker is a modular framework to automate a set of performance benchmarks, mostly for deep learning.

Run

python3 -m benchmarker --mode=training --framework=pytorch --problem=resnet50 --problem_size=32 --batch_size=4

Motivation

various devices, frameworks und underlying software stacks, network architectures etc.

Installation

Clone, install required packages for example by running

git clone --recursive https://github.com/undertherain/benchmarker.git

pip3 install [--user] -r requirements.txt

Contributors

The original version was developed in 2017 by Aleksandr Drozd. Since then to the project contributed (in alphabetical order)

• Zhengyang Bai
• Kevin Brown
• Mateusz Bysiek
• Aleksandr Drozd
• Artur Podobas
• Shweta Salaria
• Emil Vatai