diff options
author | 2019-02-04 16:55:44 +0000 | |
---|---|---|
committer | 2019-02-04 16:55:44 +0000 | |
commit | 76c648e7a477ffb2a882ad5ffe523269bd9a3f6a (patch) | |
tree | 29d319d598650bab04e4f58e5e8769567e33091e /lib/libcxx/utils/google-benchmark/tools | |
parent | Import libc++abi 7.0.1. (diff) | |
download | wireguard-openbsd-76c648e7a477ffb2a882ad5ffe523269bd9a3f6a.tar.xz wireguard-openbsd-76c648e7a477ffb2a882ad5ffe523269bd9a3f6a.zip |
Import libc++ 7.0.1.
Diffstat (limited to 'lib/libcxx/utils/google-benchmark/tools')
9 files changed, 1008 insertions, 34 deletions
diff --git a/lib/libcxx/utils/google-benchmark/tools/compare.py b/lib/libcxx/utils/google-benchmark/tools/compare.py new file mode 100755 index 00000000000..d27e24b3492 --- /dev/null +++ b/lib/libcxx/utils/google-benchmark/tools/compare.py @@ -0,0 +1,373 @@ +#!/usr/bin/env python + +""" +compare.py - versatile benchmark output compare tool +""" + +import argparse +from argparse import ArgumentParser +import sys +import gbench +from gbench import util, report +from gbench.util import * + + +def check_inputs(in1, in2, flags): + """ + Perform checking on the user provided inputs and diagnose any abnormalities + """ + in1_kind, in1_err = classify_input_file(in1) + in2_kind, in2_err = classify_input_file(in2) + output_file = find_benchmark_flag('--benchmark_out=', flags) + output_type = find_benchmark_flag('--benchmark_out_format=', flags) + if in1_kind == IT_Executable and in2_kind == IT_Executable and output_file: + print(("WARNING: '--benchmark_out=%s' will be passed to both " + "benchmarks causing it to be overwritten") % output_file) + if in1_kind == IT_JSON and in2_kind == IT_JSON and len(flags) > 0: + print("WARNING: passing optional flags has no effect since both " + "inputs are JSON") + if output_type is not None and output_type != 'json': + print(("ERROR: passing '--benchmark_out_format=%s' to 'compare.py`" + " is not supported.") % output_type) + sys.exit(1) + + +def create_parser(): + parser = ArgumentParser( + description='versatile benchmark output compare tool') + + utest = parser.add_argument_group() + utest.add_argument( + '--no-utest', + dest='utest', + default=True, + action="store_false", + help="The tool can do a two-tailed Mann-Whitney U test with the null hypothesis that it is equally likely that a randomly selected value from one sample will be less than or greater than a randomly selected value from a second sample.\nWARNING: requires **LARGE** (no less than {}) number of repetitions to be meaningful!\nThe test is being done by default, if at least {} repetitions were done.\nThis option can disable the U Test.".format(report.UTEST_OPTIMAL_REPETITIONS, report.UTEST_MIN_REPETITIONS)) + alpha_default = 0.05 + utest.add_argument( + "--alpha", + dest='utest_alpha', + default=alpha_default, + type=float, + help=("significance level alpha. if the calculated p-value is below this value, then the result is said to be statistically significant and the null hypothesis is rejected.\n(default: %0.4f)") % + alpha_default) + + subparsers = parser.add_subparsers( + help='This tool has multiple modes of operation:', + dest='mode') + + parser_a = subparsers.add_parser( + 'benchmarks', + help='The most simple use-case, compare all the output of these two benchmarks') + baseline = parser_a.add_argument_group( + 'baseline', 'The benchmark baseline') + baseline.add_argument( + 'test_baseline', + metavar='test_baseline', + type=argparse.FileType('r'), + nargs=1, + help='A benchmark executable or JSON output file') + contender = parser_a.add_argument_group( + 'contender', 'The benchmark that will be compared against the baseline') + contender.add_argument( + 'test_contender', + metavar='test_contender', + type=argparse.FileType('r'), + nargs=1, + help='A benchmark executable or JSON output file') + parser_a.add_argument( + 'benchmark_options', + metavar='benchmark_options', + nargs=argparse.REMAINDER, + help='Arguments to pass when running benchmark executables') + + parser_b = subparsers.add_parser( + 'filters', help='Compare filter one with the filter two of benchmark') + baseline = parser_b.add_argument_group( + 'baseline', 'The benchmark baseline') + baseline.add_argument( + 'test', + metavar='test', + type=argparse.FileType('r'), + nargs=1, + help='A benchmark executable or JSON output file') + baseline.add_argument( + 'filter_baseline', + metavar='filter_baseline', + type=str, + nargs=1, + help='The first filter, that will be used as baseline') + contender = parser_b.add_argument_group( + 'contender', 'The benchmark that will be compared against the baseline') + contender.add_argument( + 'filter_contender', + metavar='filter_contender', + type=str, + nargs=1, + help='The second filter, that will be compared against the baseline') + parser_b.add_argument( + 'benchmark_options', + metavar='benchmark_options', + nargs=argparse.REMAINDER, + help='Arguments to pass when running benchmark executables') + + parser_c = subparsers.add_parser( + 'benchmarksfiltered', + help='Compare filter one of first benchmark with filter two of the second benchmark') + baseline = parser_c.add_argument_group( + 'baseline', 'The benchmark baseline') + baseline.add_argument( + 'test_baseline', + metavar='test_baseline', + type=argparse.FileType('r'), + nargs=1, + help='A benchmark executable or JSON output file') + baseline.add_argument( + 'filter_baseline', + metavar='filter_baseline', + type=str, + nargs=1, + help='The first filter, that will be used as baseline') + contender = parser_c.add_argument_group( + 'contender', 'The benchmark that will be compared against the baseline') + contender.add_argument( + 'test_contender', + metavar='test_contender', + type=argparse.FileType('r'), + nargs=1, + help='The second benchmark executable or JSON output file, that will be compared against the baseline') + contender.add_argument( + 'filter_contender', + metavar='filter_contender', + type=str, + nargs=1, + help='The second filter, that will be compared against the baseline') + parser_c.add_argument( + 'benchmark_options', + metavar='benchmark_options', + nargs=argparse.REMAINDER, + help='Arguments to pass when running benchmark executables') + + return parser + + +def main(): + # Parse the command line flags + parser = create_parser() + args, unknown_args = parser.parse_known_args() + if args.mode is None: + parser.print_help() + exit(1) + assert not unknown_args + benchmark_options = args.benchmark_options + + if args.mode == 'benchmarks': + test_baseline = args.test_baseline[0].name + test_contender = args.test_contender[0].name + filter_baseline = '' + filter_contender = '' + + # NOTE: if test_baseline == test_contender, you are analyzing the stdev + + description = 'Comparing %s to %s' % (test_baseline, test_contender) + elif args.mode == 'filters': + test_baseline = args.test[0].name + test_contender = args.test[0].name + filter_baseline = args.filter_baseline[0] + filter_contender = args.filter_contender[0] + + # NOTE: if filter_baseline == filter_contender, you are analyzing the + # stdev + + description = 'Comparing %s to %s (from %s)' % ( + filter_baseline, filter_contender, args.test[0].name) + elif args.mode == 'benchmarksfiltered': + test_baseline = args.test_baseline[0].name + test_contender = args.test_contender[0].name + filter_baseline = args.filter_baseline[0] + filter_contender = args.filter_contender[0] + + # NOTE: if test_baseline == test_contender and + # filter_baseline == filter_contender, you are analyzing the stdev + + description = 'Comparing %s (from %s) to %s (from %s)' % ( + filter_baseline, test_baseline, filter_contender, test_contender) + else: + # should never happen + print("Unrecognized mode of operation: '%s'" % args.mode) + parser.print_help() + exit(1) + + check_inputs(test_baseline, test_contender, benchmark_options) + + options_baseline = [] + options_contender = [] + + if filter_baseline and filter_contender: + options_baseline = ['--benchmark_filter=%s' % filter_baseline] + options_contender = ['--benchmark_filter=%s' % filter_contender] + + # Run the benchmarks and report the results + json1 = json1_orig = gbench.util.run_or_load_benchmark( + test_baseline, benchmark_options + options_baseline) + json2 = json2_orig = gbench.util.run_or_load_benchmark( + test_contender, benchmark_options + options_contender) + + # Now, filter the benchmarks so that the difference report can work + if filter_baseline and filter_contender: + replacement = '[%s vs. %s]' % (filter_baseline, filter_contender) + json1 = gbench.report.filter_benchmark( + json1_orig, filter_baseline, replacement) + json2 = gbench.report.filter_benchmark( + json2_orig, filter_contender, replacement) + + # Diff and output + output_lines = gbench.report.generate_difference_report( + json1, json2, args.utest, args.utest_alpha) + print(description) + for ln in output_lines: + print(ln) + + +import unittest + + +class TestParser(unittest.TestCase): + def setUp(self): + self.parser = create_parser() + testInputs = os.path.join( + os.path.dirname( + os.path.realpath(__file__)), + 'gbench', + 'Inputs') + self.testInput0 = os.path.join(testInputs, 'test1_run1.json') + self.testInput1 = os.path.join(testInputs, 'test1_run2.json') + + def test_benchmarks_basic(self): + parsed = self.parser.parse_args( + ['benchmarks', self.testInput0, self.testInput1]) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'benchmarks') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertFalse(parsed.benchmark_options) + + def test_benchmarks_basic_without_utest(self): + parsed = self.parser.parse_args( + ['--no-utest', 'benchmarks', self.testInput0, self.testInput1]) + self.assertFalse(parsed.utest) + self.assertEqual(parsed.utest_alpha, 0.05) + self.assertEqual(parsed.mode, 'benchmarks') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertFalse(parsed.benchmark_options) + + def test_benchmarks_basic_with_utest_alpha(self): + parsed = self.parser.parse_args( + ['--alpha=0.314', 'benchmarks', self.testInput0, self.testInput1]) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.utest_alpha, 0.314) + self.assertEqual(parsed.mode, 'benchmarks') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertFalse(parsed.benchmark_options) + + def test_benchmarks_basic_without_utest_with_utest_alpha(self): + parsed = self.parser.parse_args( + ['--no-utest', '--alpha=0.314', 'benchmarks', self.testInput0, self.testInput1]) + self.assertFalse(parsed.utest) + self.assertEqual(parsed.utest_alpha, 0.314) + self.assertEqual(parsed.mode, 'benchmarks') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertFalse(parsed.benchmark_options) + + def test_benchmarks_with_remainder(self): + parsed = self.parser.parse_args( + ['benchmarks', self.testInput0, self.testInput1, 'd']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'benchmarks') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertEqual(parsed.benchmark_options, ['d']) + + def test_benchmarks_with_remainder_after_doubleminus(self): + parsed = self.parser.parse_args( + ['benchmarks', self.testInput0, self.testInput1, '--', 'e']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'benchmarks') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertEqual(parsed.benchmark_options, ['e']) + + def test_filters_basic(self): + parsed = self.parser.parse_args( + ['filters', self.testInput0, 'c', 'd']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'filters') + self.assertEqual(parsed.test[0].name, self.testInput0) + self.assertEqual(parsed.filter_baseline[0], 'c') + self.assertEqual(parsed.filter_contender[0], 'd') + self.assertFalse(parsed.benchmark_options) + + def test_filters_with_remainder(self): + parsed = self.parser.parse_args( + ['filters', self.testInput0, 'c', 'd', 'e']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'filters') + self.assertEqual(parsed.test[0].name, self.testInput0) + self.assertEqual(parsed.filter_baseline[0], 'c') + self.assertEqual(parsed.filter_contender[0], 'd') + self.assertEqual(parsed.benchmark_options, ['e']) + + def test_filters_with_remainder_after_doubleminus(self): + parsed = self.parser.parse_args( + ['filters', self.testInput0, 'c', 'd', '--', 'f']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'filters') + self.assertEqual(parsed.test[0].name, self.testInput0) + self.assertEqual(parsed.filter_baseline[0], 'c') + self.assertEqual(parsed.filter_contender[0], 'd') + self.assertEqual(parsed.benchmark_options, ['f']) + + def test_benchmarksfiltered_basic(self): + parsed = self.parser.parse_args( + ['benchmarksfiltered', self.testInput0, 'c', self.testInput1, 'e']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'benchmarksfiltered') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.filter_baseline[0], 'c') + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertEqual(parsed.filter_contender[0], 'e') + self.assertFalse(parsed.benchmark_options) + + def test_benchmarksfiltered_with_remainder(self): + parsed = self.parser.parse_args( + ['benchmarksfiltered', self.testInput0, 'c', self.testInput1, 'e', 'f']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'benchmarksfiltered') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.filter_baseline[0], 'c') + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertEqual(parsed.filter_contender[0], 'e') + self.assertEqual(parsed.benchmark_options[0], 'f') + + def test_benchmarksfiltered_with_remainder_after_doubleminus(self): + parsed = self.parser.parse_args( + ['benchmarksfiltered', self.testInput0, 'c', self.testInput1, 'e', '--', 'g']) + self.assertTrue(parsed.utest) + self.assertEqual(parsed.mode, 'benchmarksfiltered') + self.assertEqual(parsed.test_baseline[0].name, self.testInput0) + self.assertEqual(parsed.filter_baseline[0], 'c') + self.assertEqual(parsed.test_contender[0].name, self.testInput1) + self.assertEqual(parsed.filter_contender[0], 'e') + self.assertEqual(parsed.benchmark_options[0], 'g') + + +if __name__ == '__main__': + # unittest.main() + main() + +# vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4 +# kate: tab-width: 4; replace-tabs on; indent-width 4; tab-indents: off; +# kate: indent-mode python; remove-trailing-spaces modified; diff --git a/lib/libcxx/utils/google-benchmark/tools/compare_bench.py b/lib/libcxx/utils/google-benchmark/tools/compare_bench.py index d54baaa0e8f..7bbf0d01574 100644 --- a/lib/libcxx/utils/google-benchmark/tools/compare_bench.py +++ b/lib/libcxx/utils/google-benchmark/tools/compare_bench.py @@ -39,21 +39,20 @@ def main(): parser.add_argument( 'test2', metavar='test2', type=str, nargs=1, help='A benchmark executable or JSON output file') - # FIXME this is a dummy argument which will never actually match - # any --benchmark flags but it helps generate a better usage message parser.add_argument( - 'benchmark_options', metavar='benchmark_option', nargs='*', + 'benchmark_options', metavar='benchmark_options', nargs=argparse.REMAINDER, help='Arguments to pass when running benchmark executables' ) args, unknown_args = parser.parse_known_args() # Parse the command line flags test1 = args.test1[0] test2 = args.test2[0] - if args.benchmark_options: + if unknown_args: + # should never happen print("Unrecognized positional argument arguments: '%s'" - % args.benchmark_options) + % unknown_args) exit(1) - benchmark_options = unknown_args + benchmark_options = args.benchmark_options check_inputs(test1, test2, benchmark_options) # Run the benchmarks and report the results json1 = gbench.util.run_or_load_benchmark(test1, benchmark_options) diff --git a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run1.json b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run1.json index 37faed46d13..d7ec6a9c8f6 100644 --- a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run1.json +++ b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run1.json @@ -29,6 +29,20 @@ "time_unit": "ns" }, { + "name": "BM_1PercentFaster", + "iterations": 1000, + "real_time": 100, + "cpu_time": 100, + "time_unit": "ns" + }, + { + "name": "BM_1PercentSlower", + "iterations": 1000, + "real_time": 100, + "cpu_time": 100, + "time_unit": "ns" + }, + { "name": "BM_10PercentFaster", "iterations": 1000, "real_time": 100, @@ -55,6 +69,34 @@ "real_time": 10000, "cpu_time": 10000, "time_unit": "ns" + }, + { + "name": "BM_10PercentCPUToTime", + "iterations": 1000, + "real_time": 100, + "cpu_time": 100, + "time_unit": "ns" + }, + { + "name": "BM_ThirdFaster", + "iterations": 1000, + "real_time": 100, + "cpu_time": 100, + "time_unit": "ns" + }, + { + "name": "BM_BadTimeUnit", + "iterations": 1000, + "real_time": 0.4, + "cpu_time": 0.5, + "time_unit": "s" + }, + { + "name": "BM_DifferentTimeUnit", + "iterations": 1, + "real_time": 1, + "cpu_time": 1, + "time_unit": "s" } ] -}
\ No newline at end of file +} diff --git a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run2.json b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run2.json index aed5151d392..59a5ffaca4d 100644 --- a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run2.json +++ b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test1_run2.json @@ -29,6 +29,20 @@ "time_unit": "ns" }, { + "name": "BM_1PercentFaster", + "iterations": 1000, + "real_time": 98.9999999, + "cpu_time": 98.9999999, + "time_unit": "ns" + }, + { + "name": "BM_1PercentSlower", + "iterations": 1000, + "real_time": 100.9999999, + "cpu_time": 100.9999999, + "time_unit": "ns" + }, + { "name": "BM_10PercentFaster", "iterations": 1000, "real_time": 90, @@ -45,8 +59,8 @@ { "name": "BM_100xSlower", "iterations": 1000, - "real_time": 10000, - "cpu_time": 10000, + "real_time": 1.0000e+04, + "cpu_time": 1.0000e+04, "time_unit": "ns" }, { @@ -55,6 +69,34 @@ "real_time": 100, "cpu_time": 100, "time_unit": "ns" + }, + { + "name": "BM_10PercentCPUToTime", + "iterations": 1000, + "real_time": 110, + "cpu_time": 90, + "time_unit": "ns" + }, + { + "name": "BM_ThirdFaster", + "iterations": 1000, + "real_time": 66.665, + "cpu_time": 66.664, + "time_unit": "ns" + }, + { + "name": "BM_BadTimeUnit", + "iterations": 1000, + "real_time": 0.04, + "cpu_time": 0.6, + "time_unit": "s" + }, + { + "name": "BM_DifferentTimeUnit", + "iterations": 1, + "real_time": 1, + "cpu_time": 1, + "time_unit": "ns" } ] -}
\ No newline at end of file +} diff --git a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test2_run.json b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test2_run.json new file mode 100644 index 00000000000..15bc6980304 --- /dev/null +++ b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test2_run.json @@ -0,0 +1,81 @@ +{ + "context": { + "date": "2016-08-02 17:44:46", + "num_cpus": 4, + "mhz_per_cpu": 4228, + "cpu_scaling_enabled": false, + "library_build_type": "release" + }, + "benchmarks": [ + { + "name": "BM_Hi", + "iterations": 1234, + "real_time": 42, + "cpu_time": 24, + "time_unit": "ms" + }, + { + "name": "BM_Zero", + "iterations": 1000, + "real_time": 10, + "cpu_time": 10, + "time_unit": "ns" + }, + { + "name": "BM_Zero/4", + "iterations": 4000, + "real_time": 40, + "cpu_time": 40, + "time_unit": "ns" + }, + { + "name": "Prefix/BM_Zero", + "iterations": 2000, + "real_time": 20, + "cpu_time": 20, + "time_unit": "ns" + }, + { + "name": "Prefix/BM_Zero/3", + "iterations": 3000, + "real_time": 30, + "cpu_time": 30, + "time_unit": "ns" + }, + { + "name": "BM_One", + "iterations": 5000, + "real_time": 5, + "cpu_time": 5, + "time_unit": "ns" + }, + { + "name": "BM_One/4", + "iterations": 2000, + "real_time": 20, + "cpu_time": 20, + "time_unit": "ns" + }, + { + "name": "Prefix/BM_One", + "iterations": 1000, + "real_time": 10, + "cpu_time": 10, + "time_unit": "ns" + }, + { + "name": "Prefix/BM_One/3", + "iterations": 1500, + "real_time": 15, + "cpu_time": 15, + "time_unit": "ns" + }, + { + "name": "BM_Bye", + "iterations": 5321, + "real_time": 11, + "cpu_time": 63, + "time_unit": "ns" + } + ] +} diff --git a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test3_run0.json b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test3_run0.json new file mode 100644 index 00000000000..ca793f3367e --- /dev/null +++ b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test3_run0.json @@ -0,0 +1,39 @@ +{ + "context": { + "date": "2016-08-02 17:44:46", + "num_cpus": 4, + "mhz_per_cpu": 4228, + "cpu_scaling_enabled": false, + "library_build_type": "release" + }, + "benchmarks": [ + { + "name": "BM_One", + "iterations": 1000, + "real_time": 10, + "cpu_time": 100, + "time_unit": "ns" + }, + { + "name": "BM_Two", + "iterations": 1000, + "real_time": 9, + "cpu_time": 90, + "time_unit": "ns" + }, + { + "name": "BM_Two", + "iterations": 1000, + "real_time": 8, + "cpu_time": 80, + "time_unit": "ns" + }, + { + "name": "short", + "iterations": 1000, + "real_time": 8, + "cpu_time": 80, + "time_unit": "ns" + } + ] +} diff --git a/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test3_run1.json b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test3_run1.json new file mode 100644 index 00000000000..e5cf50c7445 --- /dev/null +++ b/lib/libcxx/utils/google-benchmark/tools/gbench/Inputs/test3_run1.json @@ -0,0 +1,39 @@ +{ + "context": { + "date": "2016-08-02 17:44:46", + "num_cpus": 4, + "mhz_per_cpu": 4228, + "cpu_scaling_enabled": false, + "library_build_type": "release" + }, + "benchmarks": [ + { + "name": "BM_One", + "iterations": 1000, + "real_time": 9, + "cpu_time": 110, + "time_unit": "ns" + }, + { + "name": "BM_Two", + "iterations": 1000, + "real_time": 10, + "cpu_time": 89, + "time_unit": "ns" + }, + { + "name": "BM_Two", + "iterations": 1000, + "real_time": 7, + "cpu_time": 70, + "time_unit": "ns" + }, + { + "name": "short", + "iterations": 1000, + "real_time": 8, + "cpu_time": 80, + "time_unit": "ns" + } + ] +} diff --git a/lib/libcxx/utils/google-benchmark/tools/gbench/report.py b/lib/libcxx/utils/google-benchmark/tools/gbench/report.py index 8f1b0fa8604..4d03a547677 100644 --- a/lib/libcxx/utils/google-benchmark/tools/gbench/report.py +++ b/lib/libcxx/utils/google-benchmark/tools/gbench/report.py @@ -1,6 +1,11 @@ """report.py - Utilities for reporting statistics about benchmark results """ import os +import re +import copy + +from scipy.stats import mannwhitneyu + class BenchmarkColor(object): def __init__(self, name, code): @@ -14,11 +19,13 @@ class BenchmarkColor(object): def __format__(self, format): return self.code + # Benchmark Colors Enumeration BC_NONE = BenchmarkColor('NONE', '') BC_MAGENTA = BenchmarkColor('MAGENTA', '\033[95m') BC_CYAN = BenchmarkColor('CYAN', '\033[96m') BC_OKBLUE = BenchmarkColor('OKBLUE', '\033[94m') +BC_OKGREEN = BenchmarkColor('OKGREEN', '\033[32m') BC_HEADER = BenchmarkColor('HEADER', '\033[92m') BC_WARNING = BenchmarkColor('WARNING', '\033[93m') BC_WHITE = BenchmarkColor('WHITE', '\033[97m') @@ -27,6 +34,10 @@ BC_ENDC = BenchmarkColor('ENDC', '\033[0m') BC_BOLD = BenchmarkColor('BOLD', '\033[1m') BC_UNDERLINE = BenchmarkColor('UNDERLINE', '\033[4m') +UTEST_MIN_REPETITIONS = 2 +UTEST_OPTIMAL_REPETITIONS = 9 # Lowest reasonable number, More is better. + + def color_format(use_color, fmt_str, *args, **kwargs): """ Return the result of 'fmt_str.format(*args, **kwargs)' after transforming @@ -66,25 +77,107 @@ def calculate_change(old_val, new_val): return float(new_val - old_val) / abs(old_val) -def generate_difference_report(json1, json2, use_color=True): +def filter_benchmark(json_orig, family, replacement=""): + """ + Apply a filter to the json, and only leave the 'family' of benchmarks. + """ + regex = re.compile(family) + filtered = {} + filtered['benchmarks'] = [] + for be in json_orig['benchmarks']: + if not regex.search(be['name']): + continue + filteredbench = copy.deepcopy(be) # Do NOT modify the old name! + filteredbench['name'] = regex.sub(replacement, filteredbench['name']) + filtered['benchmarks'].append(filteredbench) + return filtered + + +def generate_difference_report( + json1, + json2, + utest=False, + utest_alpha=0.05, + use_color=True): """ Calculate and report the difference between each test of two benchmarks runs specified as 'json1' and 'json2'. """ - first_col_width = find_longest_name(json1['benchmarks']) + 5 + assert utest is True or utest is False + first_col_width = find_longest_name(json1['benchmarks']) + def find_test(name): for b in json2['benchmarks']: if b['name'] == name: return b return None - first_line = "{:<{}s} Time CPU Old New".format( - 'Benchmark', first_col_width) + + utest_col_name = "_pvalue" + first_col_width = max( + first_col_width, + len('Benchmark')) + first_col_width += len(utest_col_name) + first_line = "{:<{}s}Time CPU Time Old Time New CPU Old CPU New".format( + 'Benchmark', 12 + first_col_width) output_strs = [first_line, '-' * len(first_line)] - for bn in json1['benchmarks']: + + last_name = None + timings_time = [[], []] + timings_cpu = [[], []] + + gen = (bn for bn in json1['benchmarks'] + if 'real_time' in bn and 'cpu_time' in bn) + for bn in gen: + fmt_str = "{}{:<{}s}{endc}{}{:+16.4f}{endc}{}{:+16.4f}{endc}{:14.0f}{:14.0f}{endc}{:14.0f}{:14.0f}" + special_str = "{}{:<{}s}{endc}{}{:16.4f}{endc}{}{:16.4f}{endc}{} {}" + + if last_name is None: + last_name = bn['name'] + if last_name != bn['name']: + if ((len(timings_time[0]) >= UTEST_MIN_REPETITIONS) and + (len(timings_time[1]) >= UTEST_MIN_REPETITIONS) and + (len(timings_cpu[0]) >= UTEST_MIN_REPETITIONS) and + (len(timings_cpu[1]) >= UTEST_MIN_REPETITIONS)): + if utest: + def get_utest_color(pval): + if pval >= utest_alpha: + return BC_FAIL + else: + return BC_OKGREEN + time_pvalue = mannwhitneyu( + timings_time[0], timings_time[1], alternative='two-sided').pvalue + cpu_pvalue = mannwhitneyu( + timings_cpu[0], timings_cpu[1], alternative='two-sided').pvalue + dsc = "U Test, Repetitions: {}".format(len(timings_cpu[0])) + dsc_color = BC_OKGREEN + if len(timings_cpu[0]) < UTEST_OPTIMAL_REPETITIONS: + dsc_color = BC_WARNING + dsc += ". WARNING: Results unreliable! {}+ repetitions recommended.".format( + UTEST_OPTIMAL_REPETITIONS) + output_strs += [color_format(use_color, + special_str, + BC_HEADER, + "{}{}".format(last_name, + utest_col_name), + first_col_width, + get_utest_color(time_pvalue), + time_pvalue, + get_utest_color(cpu_pvalue), + cpu_pvalue, + dsc_color, + dsc, + endc=BC_ENDC)] + last_name = bn['name'] + timings_time = [[], []] + timings_cpu = [[], []] + other_bench = find_test(bn['name']) if not other_bench: continue + if bn['time_unit'] != other_bench['time_unit']: + continue + def get_color(res): if res > 0.05: return BC_FAIL @@ -92,25 +185,44 @@ def generate_difference_report(json1, json2, use_color=True): return BC_WHITE else: return BC_CYAN - fmt_str = "{}{:<{}s}{endc}{}{:+9.2f}{endc}{}{:+14.2f}{endc}{:14d}{:14d}" - tres = calculate_change(bn['real_time'], other_bench['real_time']) - cpures = calculate_change(bn['cpu_time'], other_bench['cpu_time']) - output_strs += [color_format(use_color, fmt_str, - BC_HEADER, bn['name'], first_col_width, - get_color(tres), tres, get_color(cpures), cpures, - bn['cpu_time'], other_bench['cpu_time'], - endc=BC_ENDC)] + + timings_time[0].append(bn['real_time']) + timings_time[1].append(other_bench['real_time']) + timings_cpu[0].append(bn['cpu_time']) + timings_cpu[1].append(other_bench['cpu_time']) + + tres = calculate_change(timings_time[0][-1], timings_time[1][-1]) + cpures = calculate_change(timings_cpu[0][-1], timings_cpu[1][-1]) + output_strs += [color_format(use_color, + fmt_str, + BC_HEADER, + bn['name'], + first_col_width, + get_color(tres), + tres, + get_color(cpures), + cpures, + timings_time[0][-1], + timings_time[1][-1], + timings_cpu[0][-1], + timings_cpu[1][-1], + endc=BC_ENDC)] return output_strs ############################################################################### # Unit tests + import unittest + class TestReportDifference(unittest.TestCase): def load_results(self): import json - testInputs = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'Inputs') + testInputs = os.path.join( + os.path.dirname( + os.path.realpath(__file__)), + 'Inputs') testOutput1 = os.path.join(testInputs, 'test1_run1.json') testOutput2 = os.path.join(testInputs, 'test1_run2.json') with open(testOutput1, 'r') as f: @@ -121,24 +233,120 @@ class TestReportDifference(unittest.TestCase): def test_basic(self): expect_lines = [ - ['BM_SameTimes', '+0.00', '+0.00', '10', '10'], - ['BM_2xFaster', '-0.50', '-0.50', '50', '25'], - ['BM_2xSlower', '+1.00', '+1.00', '50', '100'], - ['BM_10PercentFaster', '-0.10', '-0.10', '100', '90'], - ['BM_10PercentSlower', '+0.10', '+0.10', '100', '110'], - ['BM_100xSlower', '+99.00', '+99.00', '100', '10000'], - ['BM_100xFaster', '-0.99', '-0.99', '10000', '100'], + ['BM_SameTimes', '+0.0000', '+0.0000', '10', '10', '10', '10'], + ['BM_2xFaster', '-0.5000', '-0.5000', '50', '25', '50', '25'], + ['BM_2xSlower', '+1.0000', '+1.0000', '50', '100', '50', '100'], + ['BM_1PercentFaster', '-0.0100', '-0.0100', '100', '99', '100', '99'], + ['BM_1PercentSlower', '+0.0100', '+0.0100', '100', '101', '100', '101'], + ['BM_10PercentFaster', '-0.1000', '-0.1000', '100', '90', '100', '90'], + ['BM_10PercentSlower', '+0.1000', '+0.1000', '100', '110', '100', '110'], + ['BM_100xSlower', '+99.0000', '+99.0000', + '100', '10000', '100', '10000'], + ['BM_100xFaster', '-0.9900', '-0.9900', + '10000', '100', '10000', '100'], + ['BM_10PercentCPUToTime', '+0.1000', + '-0.1000', '100', '110', '100', '90'], + ['BM_ThirdFaster', '-0.3333', '-0.3334', '100', '67', '100', '67'], + ['BM_BadTimeUnit', '-0.9000', '+0.2000', '0', '0', '0', '1'], ] json1, json2 = self.load_results() - output_lines_with_header = generate_difference_report(json1, json2, use_color=False) + output_lines_with_header = generate_difference_report( + json1, json2, use_color=False) output_lines = output_lines_with_header[2:] + print("\n") print("\n".join(output_lines_with_header)) self.assertEqual(len(output_lines), len(expect_lines)) - for i in xrange(0, len(output_lines)): + for i in range(0, len(output_lines)): + parts = [x for x in output_lines[i].split(' ') if x] + self.assertEqual(len(parts), 7) + self.assertEqual(parts, expect_lines[i]) + + +class TestReportDifferenceBetweenFamilies(unittest.TestCase): + def load_result(self): + import json + testInputs = os.path.join( + os.path.dirname( + os.path.realpath(__file__)), + 'Inputs') + testOutput = os.path.join(testInputs, 'test2_run.json') + with open(testOutput, 'r') as f: + json = json.load(f) + return json + + def test_basic(self): + expect_lines = [ + ['.', '-0.5000', '-0.5000', '10', '5', '10', '5'], + ['./4', '-0.5000', '-0.5000', '40', '20', '40', '20'], + ['Prefix/.', '-0.5000', '-0.5000', '20', '10', '20', '10'], + ['Prefix/./3', '-0.5000', '-0.5000', '30', '15', '30', '15'], + ] + json = self.load_result() + json1 = filter_benchmark(json, "BM_Z.ro", ".") + json2 = filter_benchmark(json, "BM_O.e", ".") + output_lines_with_header = generate_difference_report( + json1, json2, use_color=False) + output_lines = output_lines_with_header[2:] + print("\n") + print("\n".join(output_lines_with_header)) + self.assertEqual(len(output_lines), len(expect_lines)) + for i in range(0, len(output_lines)): + parts = [x for x in output_lines[i].split(' ') if x] + self.assertEqual(len(parts), 7) + self.assertEqual(parts, expect_lines[i]) + + +class TestReportDifferenceWithUTest(unittest.TestCase): + def load_results(self): + import json + testInputs = os.path.join( + os.path.dirname( + os.path.realpath(__file__)), + 'Inputs') + testOutput1 = os.path.join(testInputs, 'test3_run0.json') + testOutput2 = os.path.join(testInputs, 'test3_run1.json') + with open(testOutput1, 'r') as f: + json1 = json.load(f) + with open(testOutput2, 'r') as f: + json2 = json.load(f) + return json1, json2 + + def test_utest(self): + expect_lines = [] + expect_lines = [ + ['BM_One', '-0.1000', '+0.1000', '10', '9', '100', '110'], + ['BM_Two', '+0.1111', '-0.0111', '9', '10', '90', '89'], + ['BM_Two', '+0.2500', '+0.1125', '8', '10', '80', '89'], + ['BM_Two_pvalue', + '0.2207', + '0.6831', + 'U', + 'Test,', + 'Repetitions:', + '2.', + 'WARNING:', + 'Results', + 'unreliable!', + '9+', + 'repetitions', + 'recommended.'], + ['short', '+0.0000', '+0.0000', '8', '8', '80', '80'], + ] + json1, json2 = self.load_results() + output_lines_with_header = generate_difference_report( + json1, json2, True, 0.05, use_color=False) + output_lines = output_lines_with_header[2:] + print("\n") + print("\n".join(output_lines_with_header)) + self.assertEqual(len(output_lines), len(expect_lines)) + for i in range(0, len(output_lines)): parts = [x for x in output_lines[i].split(' ') if x] - self.assertEqual(len(parts), 5) self.assertEqual(parts, expect_lines[i]) if __name__ == '__main__': unittest.main() + +# vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4 +# kate: tab-width: 4; replace-tabs on; indent-width 4; tab-indents: off; +# kate: indent-mode python; remove-trailing-spaces modified; diff --git a/lib/libcxx/utils/google-benchmark/tools/strip_asm.py b/lib/libcxx/utils/google-benchmark/tools/strip_asm.py new file mode 100755 index 00000000000..9030550b43b --- /dev/null +++ b/lib/libcxx/utils/google-benchmark/tools/strip_asm.py @@ -0,0 +1,151 @@ +#!/usr/bin/env python + +""" +strip_asm.py - Cleanup ASM output for the specified file +""" + +from argparse import ArgumentParser +import sys +import os +import re + +def find_used_labels(asm): + found = set() + label_re = re.compile("\s*j[a-z]+\s+\.L([a-zA-Z0-9][a-zA-Z0-9_]*)") + for l in asm.splitlines(): + m = label_re.match(l) + if m: + found.add('.L%s' % m.group(1)) + return found + + +def normalize_labels(asm): + decls = set() + label_decl = re.compile("^[.]{0,1}L([a-zA-Z0-9][a-zA-Z0-9_]*)(?=:)") + for l in asm.splitlines(): + m = label_decl.match(l) + if m: + decls.add(m.group(0)) + if len(decls) == 0: + return asm + needs_dot = next(iter(decls))[0] != '.' + if not needs_dot: + return asm + for ld in decls: + asm = re.sub("(^|\s+)" + ld + "(?=:|\s)", '\\1.' + ld, asm) + return asm + + +def transform_labels(asm): + asm = normalize_labels(asm) + used_decls = find_used_labels(asm) + new_asm = '' + label_decl = re.compile("^\.L([a-zA-Z0-9][a-zA-Z0-9_]*)(?=:)") + for l in asm.splitlines(): + m = label_decl.match(l) + if not m or m.group(0) in used_decls: + new_asm += l + new_asm += '\n' + return new_asm + + +def is_identifier(tk): + if len(tk) == 0: + return False + first = tk[0] + if not first.isalpha() and first != '_': + return False + for i in range(1, len(tk)): + c = tk[i] + if not c.isalnum() and c != '_': + return False + return True + +def process_identifiers(l): + """ + process_identifiers - process all identifiers and modify them to have + consistent names across all platforms; specifically across ELF and MachO. + For example, MachO inserts an additional understore at the beginning of + names. This function removes that. + """ + parts = re.split(r'([a-zA-Z0-9_]+)', l) + new_line = '' + for tk in parts: + if is_identifier(tk): + if tk.startswith('__Z'): + tk = tk[1:] + elif tk.startswith('_') and len(tk) > 1 and \ + tk[1].isalpha() and tk[1] != 'Z': + tk = tk[1:] + new_line += tk + return new_line + + +def process_asm(asm): + """ + Strip the ASM of unwanted directives and lines + """ + new_contents = '' + asm = transform_labels(asm) + + # TODO: Add more things we want to remove + discard_regexes = [ + re.compile("\s+\..*$"), # directive + re.compile("\s*#(NO_APP|APP)$"), #inline ASM + re.compile("\s*#.*$"), # comment line + re.compile("\s*\.globa?l\s*([.a-zA-Z_][a-zA-Z0-9$_.]*)"), #global directive + re.compile("\s*\.(string|asciz|ascii|[1248]?byte|short|word|long|quad|value|zero)"), + ] + keep_regexes = [ + + ] + fn_label_def = re.compile("^[a-zA-Z_][a-zA-Z0-9_.]*:") + for l in asm.splitlines(): + # Remove Mach-O attribute + l = l.replace('@GOTPCREL', '') + add_line = True + for reg in discard_regexes: + if reg.match(l) is not None: + add_line = False + break + for reg in keep_regexes: + if reg.match(l) is not None: + add_line = True + break + if add_line: + if fn_label_def.match(l) and len(new_contents) != 0: + new_contents += '\n' + l = process_identifiers(l) + new_contents += l + new_contents += '\n' + return new_contents + +def main(): + parser = ArgumentParser( + description='generate a stripped assembly file') + parser.add_argument( + 'input', metavar='input', type=str, nargs=1, + help='An input assembly file') + parser.add_argument( + 'out', metavar='output', type=str, nargs=1, + help='The output file') + args, unknown_args = parser.parse_known_args() + input = args.input[0] + output = args.out[0] + if not os.path.isfile(input): + print(("ERROR: input file '%s' does not exist") % input) + sys.exit(1) + contents = None + with open(input, 'r') as f: + contents = f.read() + new_contents = process_asm(contents) + with open(output, 'w') as f: + f.write(new_contents) + + +if __name__ == '__main__': + main() + +# vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4 +# kate: tab-width: 4; replace-tabs on; indent-width 4; tab-indents: off; +# kate: indent-mode python; remove-trailing-spaces modified; |