When selecting Arm-based hardware, you may need a way to compare different systems and select a hardware configuration for your workload. Running benchmarks on different systems with different configurations is a good way to get more information about system performance.
Geekbench is a cross-platform benchmark that makes it easy to measure system performance. Preview Versions are available for Linux on Arm. You can also check the Geekbench downloads area for additional operating system options.
Geekbench provides a single-core score and a multi-core score, as well as individual performance scores for specific tests. You can use the scores to compare different systems and different configurations. A higher score is better.
Additional features are available with a purchased license .
You will need a local Arm platform or an
Arm based instance from a cloud service provider. Geekbench 5 requires
Ubuntu 16.04 LTS or later and Geekbench 6 requires
Ubuntu 18.04 LTS or later.
Both Geekbench 5 and Geekbench 6 binaries are available to download. The instructions below are for the Geekbench 6 preview version.
wget to use for the download:
sudo apt install -y wget
tar -xf Geekbench-6.0.2-LinuxARMPreview.tar.gz
There is no need for a browser or Linux desktop to run the benchmark. The Linux command line is all you need.
A number of single-core and multi-core tests are run. When complete, Geekbench uploads the results automatically and provides a link to the results. The
<id> is a number for your run.
Uploading results to the Geekbench Browser. This could take a minute or two
depending on the speed of your internet connection.
Upload succeeded. Visit the following link and view your results online:
You can create an account on Geekbench browser and save the results from your runs. This makes it easy to run Geekbench on a variety of systems and see your results together and compare them. You can also add notes to the results to help remember information about each run.
<key> will be unique numbers for your run.
Visit the following link and add this result to your profile: