Settings¶

Functions¶

Function	What it does
`fastplot.warm(processes=None)`	Compile, or load from the cache, every kernel. Run once after installing.
`fastplot.set_device(name)`	`'auto'`, `'cpu'` or `'gpu'` for every figure from now on
`fastplot.get_device()`	The current device setting
`fastplot.gpu_info()`	Whether a GPU is usable, its name, and why not if it isn't
`fastplot.tune(save=True)`	Measure the best thread count and GPU threshold on this machine
`fastplot.hardware()`	What fastplot found about the machine, and where each setting came from
`fastplot.frames(fn, items)`	Draw many figures at once on threads (Many frames)

Command line¶

python -m fastplot warm        # compile every kernel, in parallel
python -m fastplot tune        # measure and save this machine's settings
python -m fastplot hardware    # show what was found

Environment variables¶

Variable	Effect
`FASTPLOT_THREADS`	Drawing threads, overriding `tune()` and the core count
`FASTPLOT_DEVICE`	`auto`, `cpu` or `gpu`
`FASTPLOT_GPU_WORK`	The `auto` device's GPU threshold, in pixels plus grid points
`FASTPLOT_PARK`	`0` leaves numba's idle threads spinning between draws
`FASTPLOT_PRELOAD`	`0` stops the background loading of numba at import
`FASTPLOT_CACHE_DIR`	Where `tune()` keeps its results
`FASTPLOT_DATA_DIR`	Where Natural Earth's 10 m layers are downloaded to

numba's own variables apply too, for example NUMBA_NUM_THREADS to cap its thread pool, or NUMBA_CACHE_DIR when the install directory isn't writable.