Name: vega-benchmarks
Owner: Vega
Description: Scripts to benchmark Vega's performance
Created: 2015-07-20 13:26:07.0
Updated: 2016-07-19 13:48:37.0
Pushed: 2016-05-21 15:41:23.0
Homepage: null
Size: 4776
Language: HTML
GitHub Committers
| User | Most Recent Commit | # Commits |
|---|
Other Committers
| User | Most Recent Commit | # Commits |
|---|
Scripts to benchmark Vega's performance.
Average performance of rendering (non-interactive) streaming visualizations: (top-bottom) scatterplot, parallel coordinates, and trellis plot; (left-right) initialization time, average frame time, and average frame rate. Dashed lines indicate the threshold of interactive updates.
Average frame rates for three interactive visualizations: (left-right) brushing and linking on a scatterplot matrix; brushing and linking on an overview+detail visualization; panning and zooming on a scatterplot. Dashed lines indicate the threshold of interactive updates.
e: ./index.js <visualization> [options]
ons:
benchmark Benchmark operation to run. [string] [required]
renderer Vega renderer (canvas or svg). [string] [default: "canvas"]
Number of repetitions to perform. [default: 10]
Number of data tuples. [default: 100]
Number of data categories. [default: 50]
% of data tuples to stream. [default: 0.01]
help Show help [boolean]
Run npm install in the benchmarks directory, and then npm start.
Individual benchmarks can be run via the index.js script. See usage help text above.
Benchmark suites, as reported in the paper, are available via the
streaming.sh and interaction.sh bash scripts.
Results concatenated against existing results found in the results/raw folder. By default, we included the results reported in the paper. Clearing these results will generate new files instead.
A script to summarize raw results is provided for convenience (scripts/summary.js).