From: Dale Curtis <[email protected]> This patch adds the unixbench benchmark suite v5 to autotest.
Changes from v2: - self.err is a list that will hold indexes of the faulty lines of output, allowing better and more descriptive error messages. - Store raw, unprocessed benchmark output on the test resultsdir, for people wanting to use their own postprocessing scripts. Signed-off-by: Dale Curtis <[email protected]> --- client/tests/unixbench5/Makefile.patch | 11 ++ client/tests/unixbench5/control | 26 ++++ client/tests/unixbench5/unixbench5.py | 251 ++++++++++++++++++++++++++++++++ 3 files changed, 288 insertions(+), 0 deletions(-) create mode 100644 client/tests/unixbench5/Makefile.patch create mode 100644 client/tests/unixbench5/control create mode 100644 client/tests/unixbench5/unixbench5.py diff --git a/client/tests/unixbench5/Makefile.patch b/client/tests/unixbench5/Makefile.patch new file mode 100644 index 0000000..27ac225 --- /dev/null +++ b/client/tests/unixbench5/Makefile.patch @@ -0,0 +1,11 @@ +--- Makefile.bak 2011-01-14 10:45:12.000000000 -0800 ++++ Makefile 2011-01-14 10:46:54.000000000 -0800 +@@ -52,7 +52,7 @@ + # COMPILER CONFIGURATION: Set "CC" to the name of the compiler to use + # to build the binary benchmarks. You should also set "$cCompiler" in the + # Run script to the name of the compiler you want to test. +-CC=gcc ++CC?=gcc + + # OPTIMISATION SETTINGS: + diff --git a/client/tests/unixbench5/control b/client/tests/unixbench5/control new file mode 100644 index 0000000..862a521 --- /dev/null +++ b/client/tests/unixbench5/control @@ -0,0 +1,26 @@ +NAME = 'Unix Bench 5' +AUTHOR = '[email protected]' +TIME = 'MEDIUM' +PURPOSE = 'Measure system level performance.' +CRITERIA = 'This test is a benchmark.' +TEST_CLASS = 'Kernel' +TEST_CATEGORY = 'Benchmark' +TEST_TYPE = 'client' +DOC = """ +This test measure system wide performance by running the following tests: + - Dhrystone - focuses on string handling. + - Whetstone - measure floating point operations. + - Execl Throughput - measure the number of execl calls per second. + - File Copy + - Pipe throughput + - Pipe-based context switching + - Process creation - number of times a process can fork and reap + - Shell Scripts - number of times a process can start and reap a script + - System Call Overhead - estimates the cost of entering and leaving the + kernel. + +For more information visit: +http://code.google.com/p/byte-unixbench/ +""" + +job.run_test('unixbench5') diff --git a/client/tests/unixbench5/unixbench5.py b/client/tests/unixbench5/unixbench5.py new file mode 100644 index 0000000..46d176e --- /dev/null +++ b/client/tests/unixbench5/unixbench5.py @@ -0,0 +1,251 @@ +import os, re +from autotest_lib.client.bin import test, utils +from autotest_lib.client.common_lib import error + + +class unixbench5(test.test): + """ + This test measure system wide performance by running the following tests: + - Dhrystone - focuses on string handling. + - Whetstone - measure floating point operations. + - Execl Throughput - measure the number of execl calls per second. + - File Copy + - Pipe throughput + - Pipe-based context switching + - Process creation - number of times a process can fork and reap + - Shell Scripts - number of times a process can start and reap a script + - System Call Overhead - estimates the cost of entering and leaving the + kernel. + + @see: http://code.google.com/p/byte-unixbench/ + @author: Dale Curtis <[email protected]> + """ + version = 1 + + + def initialize(self): + self.job.require_gcc() + self.err = [] + + + def setup(self, tarball='unixbench-5.1.3.tgz'): + """ + Compiles unixbench. + + @tarball: Path or URL to a unixbench tarball + @see: http://byte-unixbench.googlecode.com/files/unixbench-5.1.3.tgz + """ + tarball = utils.unmap_url(self.bindir, tarball, self.tmpdir) + utils.extract_tarball_to_dir(tarball, self.srcdir) + os.chdir(self.srcdir) + + utils.system('patch -p0 < ../Makefile.patch') + utils.make() + + + def run_once(self, args=''): + vars = 'UB_TMPDIR="%s" UB_RESULTDIR="%s"' % (self.tmpdir, + self.resultsdir) + os.chdir(self.srcdir) + self.report_data = utils.system_output(vars + ' ./Run ' + args) + self.results_path = os.path.join(self.resultsdir, + 'raw_output_%s' % self.iteration) + utils.open_write_close(self.results_path, self.report_data) + + + def cleanup(self): + """ + Check error index list and throw TestError if necessary. + """ + if self.err: + e_msg = ("No measured results for output lines: %s\nOutput:%s" % + (" ".join(self.err), self.report_data)) + raise error.TestError(e_msg) + + + def process_section(self, section, suffix): + keyval = {} + subsections = section.split('\n\n') + + if len(subsections) < 3: + raise error.TestError('Invalid output format. Unable to parse') + + # Process the subsection containing performance results first. + for index, line in enumerate(subsections[1].strip().split('\n')): + # Look for problems first. + if re.search('no measured results', line, flags=re.IGNORECASE): + self.err.append(str(index + 1)) + + # Every performance result line ends with 6 values, with the sixth + # being the actual result. Make sure there are at least that words + # in the line before processing. + words = line.lower().split() + if len(words) >= 6: + key = re.sub('\W', '', '_'.join(words[:-6])) + keyval[key + suffix] = words[-6] + + # The final score should be the last item in the third subsection. + keyval['score' + suffix] = subsections[2].strip().split()[-1] + + self.write_perf_keyval(keyval) + + + def postprocess_iteration(self): + # Break up sections around dividing lines. + sections = self.report_data.split('-'*72) + + # First section is junk to us, second has results for single CPU run. + if len(sections) > 1: + self.process_section(section=sections[1], suffix='') + + # Only machines with > 1 CPU will have a 3rd section. + if len(sections) > 2: + self.process_section(section=sections[2], suffix='_multi') + else: + raise error.TestError('Invalid output format. Unable to parse') + + +""" Here is a sample output: + + # # # # # # # ##### ###### # # #### # # + # # ## # # # # # # # ## # # # # # + # # # # # # ## ##### ##### # # # # ###### + # # # # # # ## # # # # # # # # # + # # # ## # # # # # # # ## # # # # + #### # # # # # ##### ###### # # #### # # + + Version 5.1.2 Based on the Byte Magazine Unix Benchmark + + Multi-CPU version Version 5 revisions by Ian Smith, + Sunnyvale, CA, USA + December 22, 2007 johantheghost at yahoo period com + + +1 x Dhrystone 2 using register variables 1 2 3 4 5 6 7 8 9 10 + +1 x Double-Precision Whetstone 1 2 3 4 5 6 7 8 9 10 + +1 x Execl Throughput 1 2 3 + +1 x File Copy 1024 bufsize 2000 maxblocks 1 2 3 + +1 x File Copy 256 bufsize 500 maxblocks 1 2 3 + +1 x File Copy 4096 bufsize 8000 maxblocks 1 2 3 + +1 x Pipe Throughput 1 2 3 4 5 6 7 8 9 10 + +1 x Pipe-based Context Switching 1 2 3 4 5 6 7 8 9 10 + +1 x Process Creation 1 2 3 + +1 x System Call Overhead 1 2 3 4 5 6 7 8 9 10 + +1 x Shell Scripts (1 concurrent) 1 2 3 + +1 x Shell Scripts (8 concurrent) 1 2 3 + +2 x Dhrystone 2 using register variables 1 2 3 4 5 6 7 8 9 10 + +2 x Double-Precision Whetstone 1 2 3 4 5 6 7 8 9 10 + +2 x Execl Throughput 1 2 3 + +2 x File Copy 1024 bufsize 2000 maxblocks 1 2 3 + +2 x File Copy 256 bufsize 500 maxblocks 1 2 3 + +2 x File Copy 4096 bufsize 8000 maxblocks 1 2 3 + +2 x Pipe Throughput 1 2 3 4 5 6 7 8 9 10 + +2 x Pipe-based Context Switching 1 2 3 4 5 6 7 8 9 10 + +2 x Process Creation 1 2 3 + +2 x System Call Overhead 1 2 3 4 5 6 7 8 9 10 + +2 x Shell Scripts (1 concurrent) 1 2 3 + +2 x Shell Scripts (8 concurrent) 1 2 3 + +======================================================================== + BYTE UNIX Benchmarks (Version 5.1.2) + + System: localhost: GNU/Linux + OS: GNU/Linux -- 2.6.32.26+drm33.12 -- #1 SMP Wed Jan 12 16:16:05 PST 2011 + Machine: i686 (GenuineIntel) + Language: en_US.utf8 (charmap=, collate=) + CPU 0: Intel(R) Atom(TM) CPU N455 @ 1.66GHz (3325.2 bogomips) + Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT + CPU 1: Intel(R) Atom(TM) CPU N455 @ 1.66GHz (3325.0 bogomips) + Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT + 14:11:59 up 1 day, 1:10, 0 users, load average: 0.47, 0.48, 0.51; runlevel + +------------------------------------------------------------------------ +Benchmark Run: Fri Jan 14 2011 14:11:59 - 14:41:26 +2 CPUs in system; running 1 parallel copy of tests + +Dhrystone 2 using register variables 2264000.6 lps (10.0 s, 7 samples) +Double-Precision Whetstone 507.0 MWIPS (10.1 s, 7 samples) +Execl Throughput 796.7 lps (30.0 s, 2 samples) +File Copy 1024 bufsize 2000 maxblocks 110924.1 KBps (30.1 s, 2 samples) +File Copy 256 bufsize 500 maxblocks 32600.5 KBps (30.1 s, 2 samples) +File Copy 4096 bufsize 8000 maxblocks 284236.5 KBps (30.0 s, 2 samples) +Pipe Throughput 301672.5 lps (10.0 s, 7 samples) +Pipe-based Context Switching 29475.3 lps (10.0 s, 7 samples) +Process Creation 3124.6 lps (30.0 s, 2 samples) +Shell Scripts (1 concurrent) 1753.0 lpm (60.0 s, 2 samples) +Shell Scripts (8 concurrent) 305.9 lpm (60.1 s, 2 samples) +System Call Overhead 592781.7 lps (10.0 s, 7 samples) + +System Benchmarks Index Values BASELINE RESULT INDEX +Dhrystone 2 using register variables 116700.0 2264000.6 194.0 +Double-Precision Whetstone 55.0 507.0 92.2 +Execl Throughput 43.0 796.7 185.3 +File Copy 1024 bufsize 2000 maxblocks 3960.0 110924.1 280.1 +File Copy 256 bufsize 500 maxblocks 1655.0 32600.5 197.0 +File Copy 4096 bufsize 8000 maxblocks 5800.0 284236.5 490.1 +Pipe Throughput 12440.0 301672.5 242.5 +Pipe-based Context Switching 4000.0 29475.3 73.7 +Process Creation 126.0 3124.6 248.0 +Shell Scripts (1 concurrent) 42.4 1753.0 413.4 +Shell Scripts (8 concurrent) 6.0 305.9 509.8 +System Call Overhead 15000.0 592781.7 395.2 + ======== +System Benchmarks Index Score 238.0 + +------------------------------------------------------------------------ +Benchmark Run: Fri Jan 14 2011 14:41:26 - 15:09:23 +2 CPUs in system; running 2 parallel copies of tests + +Dhrystone 2 using register variables 3411919.6 lps (10.0 s, 7 samples) +Double-Precision Whetstone 964.3 MWIPS (10.1 s, 7 samples) +Execl Throughput 2053.5 lps (30.0 s, 2 samples) +File Copy 1024 bufsize 2000 maxblocks 158308.0 KBps (30.0 s, 2 samples) +File Copy 256 bufsize 500 maxblocks 46249.5 KBps (30.0 s, 2 samples) +File Copy 4096 bufsize 8000 maxblocks 389881.9 KBps (30.0 s, 2 samples) +Pipe Throughput 410193.1 lps (10.0 s, 7 samples) +Pipe-based Context Switching 113780.0 lps (10.0 s, 7 samples) +Process Creation 7609.0 lps (30.0 s, 2 samples) +Shell Scripts (1 concurrent) 2355.0 lpm (60.0 s, 2 samples) +Shell Scripts (8 concurrent) 308.1 lpm (60.2 s, 2 samples) +System Call Overhead 1057063.2 lps (10.0 s, 7 samples) + +System Benchmarks Index Values BASELINE RESULT INDEX +Dhrystone 2 using register variables 116700.0 3411919.6 292.4 +Double-Precision Whetstone 55.0 964.3 175.3 +Execl Throughput 43.0 2053.5 477.6 +File Copy 1024 bufsize 2000 maxblocks 3960.0 158308.0 399.8 +File Copy 256 bufsize 500 maxblocks 1655.0 46249.5 279.5 +File Copy 4096 bufsize 8000 maxblocks 5800.0 389881.9 672.2 +Pipe Throughput 12440.0 410193.1 329.7 +Pipe-based Context Switching 4000.0 113780.0 284.5 +Process Creation 126.0 7609.0 603.9 +Shell Scripts (1 concurrent) 42.4 2355.0 555.4 +Shell Scripts (8 concurrent) 6.0 308.1 513.5 +System Call Overhead 15000.0 1057063.2 704.7 + ======== +System Benchmarks Index Score 407.4 + +""" -- 1.7.3.4 _______________________________________________ Autotest mailing list [email protected] http://test.kernel.org/cgi-bin/mailman/listinfo/autotest
