llg3d.benchmarks.efficiency¶

Compare the efficiency of different solver for different DOMAIN_SIZES.

Four modes are available:

run: run the benchmark and save results to a CSV file
report: print a results table from a CSV file
plot: plot the results from a CSV file
compare: plot efficiency comparison from various CSV files

$ llg3d.bench.efficiency --help
usage: llg3d.bench.efficiency [-h] {run,report,plot,compare} ...

Compare the efficiency of different solver for different :attr:`DOMAIN_SIZES`.
Four modes are available: - `run`: run the benchmark and save results to a CSV
file - `report`: print a results table from a CSV file - `plot`: plot the
results from a CSV file - `compare`: plot efficiency comparison from various
CSV files .. command-output:: llg3d.bench.efficiency --help

positional arguments:
  {run,report,plot,compare}
    run                 Run the benchmark and save the CSV file
    report              Print a results table from a CSV file
    plot                Plot the graph from the CSV file
    compare             Plot efficiency comparison from various CSV files

options:
  -h, --help            show this help message and exit

Module Attributes

DOMAIN_SIZES

Jx values to benchmark

Functions

`compare_csv_files`(filepaths, solver[, show])	Compare efficiency from multiple CSV files by plotting them together.
`get_cpu_gpu_legend`()	Get a legend string with CPU and GPU names.
`get_cpu_name`()	Get the CPU name of the current machine.
`get_gpu_name`()	Get the name of the OpenCL device to use for the benchmark.
`get_num_iterations`(Jx)	Get the number of iterations for the benchmark based on domain size.
`load_csv`(filepath)	Load benchmark results from a CSV file.
`main`()	Main function to run the benchmark, plot, or report results.
`plot`(results[, show, legend, errorbars, stds])	Plot the benchmark results.
`results_as_table`(results[, stds, legend])	Format the benchmark results as a table.
`run_benchmark`(nproc[, precision, repeats])	Run the benchmark for different solvers and domain sizes.
`save_as_csv`(results, filepath[, stds, legend])	Save the benchmark results as a CSV file.

DOMAIN_SIZES: tuple[int, ...] = (128, 256, 512, 1024, 2048, 4096)¶: Jx values to benchmark

get_num_iterations(Jx)[source]¶

Get the number of iterations for the benchmark based on domain size.

Parameters:: Jx (int)
Return type:: int

get_gpu_name()[source]¶

Get the name of the OpenCL device to use for the benchmark.

Return type:: str

get_cpu_name()[source]¶

Get the CPU name of the current machine.

Return type:: str | None

get_cpu_gpu_legend()[source]¶

Get a legend string with CPU and GPU names.

Return type:: str

run_benchmark(nproc, precision='single', repeats=1)[source]¶

Run the benchmark for different solvers and domain sizes.

Parameters:

nproc (int) – Number of MPI processes to use for the MPI solver.
precision (str) – Precision of the simulation (“single” or “double”).
repeats (int) – Number of times to repeat each measurement. If >1, function returns both means and std-devs per solver/domain.

Returns:

two mappings from display-name (e.g. ‘NumPy (1 CPU core)’): to lists of mean and std values per domain size. If repeats == 1, stds is None.

Return type:

(means, stds)

save_as_csv(results, filepath, stds=None, legend=None)[source]¶

Save the benchmark results as a CSV file.

Parameters:

results (dict[str, list[float]]) – The benchmark results.
filepath (Path) – The output CSV filepath.
stds (dict[str, list[float]] | None) – Optional standard deviations for the results.
legend (str | None) – An optional legend string to include as metadata.

Return type:

None

results_as_table(results, stds=None, legend='')[source]¶

Format the benchmark results as a table.

Add an acceleration column showing speedup relative to the NumPy solver.

Parameters:

results (dict[str, list[float]]) – The benchmark results.
stds (dict[str, list[float]] | None) – Optional standard deviations for the results.
legend (str | None) – An optional legend string to include above the table.

Returns:

A string representing the results in table format.

Return type:

str

plot(results, show=False, legend=None, errorbars=False, stds=None)[source]¶

Plot the benchmark results.

Parameters:

results (dict[str, list[float]]) – The benchmark results.
show (bool) – Whether to display the plot interactively .
legend (str | None) – An optional legend string to include in the plot title.
errorbars (bool) – Whether to plot error bars using stds if provided.
stds (dict[str, list[float]] | None) – Optional standard deviations for the results.

Return type:

None

load_csv(filepath)[source]¶

Load benchmark results from a CSV file.

Returns results keyed by display-name (same format as run_benchmark output).

Parameters:: filepath (Path) – The CSV filepath to load.
Returns:: A tuple (results, stds, metadata) where results is the benchmark results, stds are the standard deviations if present (else None), and metadata is any metadata extracted from the CSV file.
Return type:: tuple[dict[str, list[float]], dict[str, list[float]] | None, dict]

compare_csv_files(filepaths, solver, show=False)[source]¶

Compare efficiency from multiple CSV files by plotting them together.

Parameters:

filepaths (list[Path]) – List of CSV filepaths to compare.
solver (str) – The solver to compare (“numpy”, “mpi”, “opencl”).
show (bool) – Whether to display the plot interactively.

Return type:

None

main()[source]¶: Main function to run the benchmark, plot, or report results.