Template Benchmarks

[philipc2]

Closes #959

Overview

• Creates template benchmarks for the mpas_ocean benchmark module, which can be re-used to avoid needing to always create setup, teardown and other configurations
• Templates DatasetBenchmark and GridBenchmark are used for benchmarks that require a UxDataset or Grid respectively.

[github-actions]

ASV Benchmarking

Benchmark Comparison Results

Benchmarks that have improved:

Change	Before [3f8929d]	After [26dac3d]	Ratio	Benchmark (Parameter)
-	407M	354M	0.87	mpas_ocean.Integrate.peakmem_integrate('480km')

Benchmarks that have stayed the same:

Change	Before [3f8929d]	After [26dac3d]	Ratio	Benchmark (Parameter)
	failed	failed	n/a	face_bounds.Bounds.time_bounds
	383M	383M	1.00	face_bounds.FaceBounds.peakmem_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/mpas/QU/oQU480.231010.nc'))
	385M	385M	1.00	face_bounds.FaceBounds.peakmem_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/scrip/outCSne8/outCSne8.nc'))
	384M	388M	1.01	face_bounds.FaceBounds.peakmem_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/ugrid/geoflow-small/grid.nc'))
	382M	386M	1.01	face_bounds.FaceBounds.peakmem_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/ugrid/quad-hexagon/grid.nc'))
	1.56±0.01s	1.57±0.01s	1.00	face_bounds.FaceBounds.time_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/mpas/QU/oQU480.231010.nc'))
	223±1ms	221±0.3ms	0.99	face_bounds.FaceBounds.time_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/scrip/outCSne8/outCSne8.nc'))
	2.02±0s	2.03±0.02s	1.01	face_bounds.FaceBounds.time_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/ugrid/geoflow-small/grid.nc'))
	7.51±0.2ms	7.59±0.2ms	1.01	face_bounds.FaceBounds.time_face_bounds(PosixPath('/home/runner/work/uxarray/uxarray/test/meshfiles/ugrid/quad-hexagon/grid.nc'))
	1.76±0s	1.75±0.01s	0.99	import.Imports.timeraw_import_uxarray
	646±9ms	658±10ms	1.02	mpas_ocean.ConnectivityConstruction.time_face_face_connectivity('120km')
	41.5±0.5ms	41.5±0.3ms	1.00	mpas_ocean.ConnectivityConstruction.time_face_face_connectivity('480km')
	2.05±0.09ms	1.91±0.1ms	0.93	mpas_ocean.ConnectivityConstruction.time_n_nodes_per_face('120km')
	516±20μs	521±8μs	1.01	mpas_ocean.ConnectivityConstruction.time_n_nodes_per_face('480km')
	297±0.8ns	305±3ns	1.02	mpas_ocean.ConstructTreeStructures.time_ball_tree('480km')
	816±6ns	822±5ns	1.01	mpas_ocean.ConstructTreeStructures.time_kd_tree('120km')
	283±3ns	281±3ns	0.99	mpas_ocean.ConstructTreeStructures.time_kd_tree('480km')
	399M	399M	1.00	mpas_ocean.GeoDataFrame.peakmem_to_geodataframe('120km', False)
	389M	389M	1.00	mpas_ocean.GeoDataFrame.peakmem_to_geodataframe('120km', True)
	362M	362M	1.00	mpas_ocean.GeoDataFrame.peakmem_to_geodataframe('480km', False)
	361M	361M	1.00	mpas_ocean.GeoDataFrame.peakmem_to_geodataframe('480km', True)
	1.04±0s	1.03±0s	0.99	mpas_ocean.GeoDataFrame.time_to_geodataframe('120km', False)
	57.8±0.3ms	57.6±0.3ms	1.00	mpas_ocean.GeoDataFrame.time_to_geodataframe('120km', True)
	78.7±0.6ms	77.5±0.5ms	0.99	mpas_ocean.GeoDataFrame.time_to_geodataframe('480km', False)
	5.66±0.1ms	5.73±0.07ms	1.01	mpas_ocean.GeoDataFrame.time_to_geodataframe('480km', True)
	263M	264M	1.00	mpas_ocean.Gradient.peakmem_gradient('120km')
	241M	241M	1.00	mpas_ocean.Gradient.peakmem_gradient('480km')
	2.73±0ms	2.73±0.01ms	1.00	mpas_ocean.Gradient.time_gradient('120km')
	293±3μs	290±2μs	0.99	mpas_ocean.Gradient.time_gradient('480km')
	244±3μs	232±20μs	0.95	mpas_ocean.HoleEdgeIndices.time_construct_hole_edge_indices('120km')
	126±1μs	123±0.8μs	0.98	mpas_ocean.HoleEdgeIndices.time_construct_hole_edge_indices('480km')
	370M	371M	1.00	mpas_ocean.Integrate.peakmem_integrate('120km')
	174±1ms	174±2ms	1.00	mpas_ocean.Integrate.time_integrate('120km')
	11.8±0.05ms	11.9±0.01ms	1.01	mpas_ocean.Integrate.time_integrate('480km')
	345±1ms	346±0.7ms	1.00	mpas_ocean.MatplotlibConversion.time_dataarray_to_polycollection('120km', 'exclude')
	349±4ms	348±2ms	1.00	mpas_ocean.MatplotlibConversion.time_dataarray_to_polycollection('120km', 'include')
	344±2ms	349±1ms	1.01	mpas_ocean.MatplotlibConversion.time_dataarray_to_polycollection('120km', 'split')
	22.5±0.3ms	22.8±0.2ms	1.01	mpas_ocean.MatplotlibConversion.time_dataarray_to_polycollection('480km', 'exclude')
	23.1±0.4ms	23.4±0.3ms	1.01	mpas_ocean.MatplotlibConversion.time_dataarray_to_polycollection('480km', 'include')
	22.9±0.3ms	22.4±0.5ms	0.98	mpas_ocean.MatplotlibConversion.time_dataarray_to_polycollection('480km', 'split')
	54.7±0.1ms	55.0±0.2ms	1.01	mpas_ocean.RemapDownsample.time_inverse_distance_weighted_remapping
	44.4±0.1ms	44.4±0.06ms	1.00	mpas_ocean.RemapDownsample.time_nearest_neighbor_remapping
	359±0.4ms	359±0.5ms	1.00	mpas_ocean.RemapUpsample.time_inverse_distance_weighted_remapping
	266±2ms	264±0.3ms	0.99	mpas_ocean.RemapUpsample.time_nearest_neighbor_remapping
	237M	237M	1.00	quad_hexagon.QuadHexagon.peakmem_open_dataset
	236M	236M	1.00	quad_hexagon.QuadHexagon.peakmem_open_grid
	6.67±0.1ms	6.77±0.04ms	1.01	quad_hexagon.QuadHexagon.time_open_dataset
	5.60±0.04ms	5.71±0.1ms	1.02	quad_hexagon.QuadHexagon.time_open_grid

Benchmarks that have got worse:

Change	Before [3f8929d]	After [26dac3d]	Ratio	Benchmark (Parameter)
+	1.21±0μs	1.42±0.01μs	1.17	mpas_ocean.ConstructTreeStructures.time_ball_tree('120km')

[rajeeja]

seems to work, thanks for the cleanup.
Does this also fix asv profile issue?

[philipc2]

seems to work, thanks for the cleanup. Does this also fix asv profile issue?

Unfortunately no, still trying to figure out how to call parametrized benchmarks. Might raise an issue on the asv repo