This benchmark measures the time taken to retrieve chunks from different binary formats used in bioimaging. Currently, TIFF, HDF5 and Zarr are supported.
This benchmark was included in the original OME-NGFF paper to demonstrate the benefits of next-generation file formats.
The easiest way to get started with the benchmark is by using Docker. For more reproducible results, instructions are provided on launching dedicated Amazon EC2 instances.
To get started, clone this repository locally:
git clone https://github.com/ome/bioimage-latency-benchmark.git
cd bioimage-latency-benchmark
You will likely want to adjust the parameters in .env first, then run:
./generate.sh
which will run several docker-compose commands in a row. This could take a substantial amount of time depending on your parameters.
Start the various Docker containers in the background ("detached" mode):
docker-compose up -d
Once the containers are up, run:
docker-compose run --rm upload
docker-compose run --rm benchmark -sv
This will store both the benchmarking results ('benchmark_data.json') as well as a plotted graph ('benchmark_plot.png') in the directory along with the input data.