def create_actual_table_dense_dot_sparse(self):
labels = {'dot_numpy': 'numpy (Referenz)'}
items_pro_dim = [100]
funcs_ranked_by_time = [(0.1, 'dot_numpy')]
results = {'dot_numpy': [(0.1, 0.001)]}
timings = {'dot_numpy': [0.1]}
test_data = tf.TableData(labels, items_pro_dim, funcs_ranked_by_time,
results, timings)
return tf.create_summery_tables_dense_dot_sparse_benchmark(test_data)
def create_actual_table_for_sparsity_benchmark(self):
labels = {'bsr_matrix': 'Block Sparse Row'}
items_pro_dim = [100]
funcs_ranked_by_time = [(0.1, 'bsr_matrix')]
results = {0.1: {'bsr_matrix': [(0.1, 0.001), (0.4, 0.004)]}}
timings = {'bsr_matrix': [0.1, 0.4]}
test_data = tf.TableData(labels, items_pro_dim, funcs_ranked_by_time,
results, timings)
return tf.create_summery_tables_for_sparse_matrices_benchmark(
test_data, 0.1)
Returns a dictionary with the function names as keys and aliases as values.
-------
"""
return {'dot_numpy':'Numpy x Numpy (Referenz)', 'scipy_csc_dot_numpy_with_swap':'Compressed Sparse Column x Numpy',
'scipy_csr_dot_numpy_with_swap':'Compressed Sparse Row x Numpy',
'scipy_bsr_dot_numpy_with_swap':'Block Sparse Row x Numpy'}
if __name__ == '__main__':
benchmark_timestamp = df.get_date()
results_path = RESULTS_DIRECTORY + BENCHMARK_DIRECTORY + benchmark_timestamp + "/"
functions_under_test = get_functions_under_test()
items_pro_dimension = [500, 1000, 2000, 3000, 4000, 5000]
number_of_timings_pro_function_and_matrix_dimension = 5
results = dds.run_performance_test(items_pro_dimension, number_of_timings_pro_function_and_matrix_dimension,
functions_under_test)
dds.backup_results(results_path, results, FILENAME)
timings = dds.get_timings_from_results(results)
functions_ranked_by_time = dds.rank_functions_by_performance(timings)
functions_labels = create_functions_aliases()
table_data = tf.TableData(functions_labels, items_pro_dimension, functions_ranked_by_time, results, timings)
ranked_times = [ranked_label for time, ranked_label in functions_ranked_by_time]
plot__data = PlotData(TEST_NAME, functions_labels, results, ranked_times, PLOT_X_LABEL, PLOT_Y_LABEL)
dds.persist_plots(items_pro_dimension, results_path, plot__data)
results_table = dds.create_summery_table(table_data)
dds.persist_summery_table(number_of_timings_pro_function_and_matrix_dimension, results_path, results_table, TABLE_HEADLINE)
results = run_performance_test(sparsities, sparse_matrices,
items_pro_matrix_dimension,
number_of_timings)
backup_results(results_path, results)
timings = get_timings_from_performance_test_results(results)
functions_labels = create_functions_aliases()
functions_ranked_by_performance = rank_functions_by_performance(timings)
persist_benchmark_data_to_summery_table_file(number_of_timings,
results_path)
for key in sparsities:
plot_title = create_plot_title(key)
plot_data = PlotData(plot_title, functions_labels, results,
functions_ranked_by_performance, PLOT_X_LABEL,
PLOT_Y_LABEL)
pf.plot_timing_sparse_matrices_benchmark(plot_data,
items_pro_matrix_dimension,
sparse_matrices, key,
results_path)
pf.make_reduced_plot_for_sparse_matrices_benchmark(
plot_data, items_pro_matrix_dimension, sparse_matrices, key,
results_path)
table_data = tf.TableData(functions_labels, items_pro_matrix_dimension,
functions_ranked_by_performance[key],
results, timings[key])
persist_summery_table(table_data, key, results_path)