uxarray-asv

Air Speed Velocity (ASV) benchmarking for the UXarray Python Package

Quickstart

UXarray's benchmarks are stored in the package repository (not in this one), which should first be cloned.

git clone https://github.com/UXARRAY/uxarray.git

The benchmarks and configuration files are located in the benchmarks directory.

cd benchmarks

Asv needs to be installed locally in order to run any benchmarks.

pip install asv

Step 4

asv run --quick

An example of the output (on a single benchmark file) is shown below.

· Creating environments
· Discovering benchmarks
· Running 4 total benchmarks (1 commits * 1 environments * 4 benchmarks)
[ 0.00%] · For uxarray commit 5410c4f0 <main>:
[ 0.00%] ·· Benchmarking virtualenv-py3.11-netcdf4-pip+pyfma-setuptools_scm-xarray
[12.50%] ··· quad_hexagon.QuadHexagon.peakmem_open_dataset                                                                                                               278M
[25.00%] ··· quad_hexagon.QuadHexagon.peakmem_open_grid                                                                                                                  275M
[37.50%] ··· quad_hexagon.QuadHexagon.time_open_dataset                                                                                                              29.8±0ms
[50.00%] ··· quad_hexagon.QuadHexagon.time_open_grid                                                                                                                 20.0±0ms

Profiling

asv profile benchmark_file.BenchmarkClass.benchmark_method

For parameterized benchmarks, you can profile a single parameter using the following:

todo

To visualuze the results, ...

asv profile benchmark_file.BenchmarkClass.benchmark_method --gui=snakeviz

Local Development

By default, asv will pull the changes from the most recent committ from whichever branch is indicated by the "branches" variable in the asv.conf.json file.

First, uxarray should be installed locally in editable mode.

pip install -e .

To run benchmarks using the local changes, we can pass through the -python=same parameter, which will use the current enviroment and not create a new one.

asv run --python=same

Submitting Benchmark Results

Coming soon!