* [PATCH v3 1/4] scripts/simplebench: add simplebench.py
2020-02-28 7:19 [PATCH v3 0/4] benchmark util Vladimir Sementsov-Ogievskiy
@ 2020-02-28 7:19 ` Vladimir Sementsov-Ogievskiy
2020-02-28 13:03 ` Aleksandar Markovic
2020-02-28 7:19 ` [PATCH v3 2/4] scripts/simplebench: add qemu/bench_block_job.py Vladimir Sementsov-Ogievskiy
` (2 subsequent siblings)
3 siblings, 1 reply; 13+ messages in thread
From: Vladimir Sementsov-Ogievskiy @ 2020-02-28 7:19 UTC (permalink / raw)
To: qemu-devel
Cc: kwolf, vsementsov, ehabkost, qemu-block, stefanha, mreitz, crosa,
den, jsnow
Add simple benchmark table creator.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
---
scripts/simplebench/simplebench.py | 128 +++++++++++++++++++++++++++++
1 file changed, 128 insertions(+)
create mode 100644 scripts/simplebench/simplebench.py
diff --git a/scripts/simplebench/simplebench.py b/scripts/simplebench/simplebench.py
new file mode 100644
index 0000000000..59e7314ff6
--- /dev/null
+++ b/scripts/simplebench/simplebench.py
@@ -0,0 +1,128 @@
+#!/usr/bin/env python
+#
+# Simple benchmarking framework
+#
+# Copyright (c) 2019 Virtuozzo International GmbH.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+
+def bench_one(test_func, test_env, test_case, count=5, initial_run=True):
+ """Benchmark one test-case
+
+ test_func -- benchmarking function with prototype
+ test_func(env, case), which takes test_env and test_case
+ arguments and returns {'seconds': int} (which is benchmark
+ result) on success and {'error': str} on error. Returned
+ dict may contain any other additional fields.
+ test_env -- test environment - opaque first argument for test_func
+ test_case -- test case - opaque second argument for test_func
+ count -- how many times to call test_func, to calculate average
+ initial_run -- do initial run of test_func, which don't get into result
+
+ Returns dict with the following fields:
+ 'runs': list of test_func results
+ 'average': average seconds per run (exists only if at least one run
+ succeeded)
+ 'delta': maximum delta between test_func result and the average
+ (exists only if at least one run succeeded)
+ 'n-failed': number of failed runs (exists only if at least one run
+ failed)
+ """
+ if initial_run:
+ print(' #initial run:')
+ print(' ', test_func(test_env, test_case))
+
+ runs = []
+ for i in range(count):
+ print(' #run {}'.format(i+1))
+ res = test_func(test_env, test_case)
+ print(' ', res)
+ runs.append(res)
+
+ result = {'runs': runs}
+
+ successed = [r for r in runs if ('seconds' in r)]
+ if successed:
+ avg = sum(r['seconds'] for r in successed) / len(successed)
+ result['average'] = avg
+ result['delta'] = max(abs(r['seconds'] - avg) for r in successed)
+
+ if len(successed) < count:
+ result['n-failed'] = count - len(successed)
+
+ return result
+
+
+def ascii_one(result):
+ """Return ASCII representation of bench_one() returned dict."""
+ if 'average' in result:
+ s = '{:.2f} +- {:.2f}'.format(result['average'], result['delta'])
+ if 'n-failed' in result:
+ s += '\n({} failed)'.format(result['n-failed'])
+ return s
+ else:
+ return 'FAILED'
+
+
+def bench(test_func, test_envs, test_cases, *args, **vargs):
+ """Fill benchmark table
+
+ test_func -- benchmarking function, see bench_one for description
+ test_envs -- list of test environments, see bench_one
+ test_cases -- list of test cases, see bench_one
+ args, vargs -- additional arguments for bench_one
+
+ Returns dict with the following fields:
+ 'envs': test_envs
+ 'cases': test_cases
+ 'tab': filled 2D array, where cell [i][j] is bench_one result for
+ test_cases[i] for test_envs[j] (i.e., rows are test cases and
+ columns are test environments)
+ """
+ tab = {}
+ results = {
+ 'envs': test_envs,
+ 'cases': test_cases,
+ 'tab': tab
+ }
+ n = 1
+ n_tests = len(test_envs) * len(test_cases)
+ for env in test_envs:
+ for case in test_cases:
+ print('Testing {}/{}: {} :: {}'.format(n, n_tests,
+ env['id'], case['id']))
+ if case['id'] not in tab:
+ tab[case['id']] = {}
+ tab[case['id']][env['id']] = bench_one(test_func, env, case,
+ *args, **vargs)
+ n += 1
+
+ print('Done')
+ return results
+
+
+def ascii(results):
+ """Return ASCII representation of bench() returned dict."""
+ from tabulate import tabulate
+
+ tab = [[""] + [c['id'] for c in results['envs']]]
+ for case in results['cases']:
+ row = [case['id']]
+ for env in results['envs']:
+ row.append(ascii_one(results['tab'][case['id']][env['id']]))
+ tab.append(row)
+
+ return tabulate(tab)
--
2.21.0
^ permalink raw reply related [flat|nested] 13+ messages in thread
* Re: [PATCH v3 1/4] scripts/simplebench: add simplebench.py
2020-02-28 7:19 ` [PATCH v3 1/4] scripts/simplebench: add simplebench.py Vladimir Sementsov-Ogievskiy
@ 2020-02-28 13:03 ` Aleksandar Markovic
2020-02-28 13:48 ` Vladimir Sementsov-Ogievskiy
0 siblings, 1 reply; 13+ messages in thread
From: Aleksandar Markovic @ 2020-02-28 13:03 UTC (permalink / raw)
To: Vladimir Sementsov-Ogievskiy
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
On Fri, Feb 28, 2020 at 8:19 AM Vladimir Sementsov-Ogievskiy
<vsementsov@virtuozzo.com> wrote:
>
> Add simple benchmark table creator.
>
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> ---
> scripts/simplebench/simplebench.py | 128 +++++++++++++++++++++++++++++
> 1 file changed, 128 insertions(+)
> create mode 100644 scripts/simplebench/simplebench.py
>
> diff --git a/scripts/simplebench/simplebench.py b/scripts/simplebench/simplebench.py
> new file mode 100644
> index 0000000000..59e7314ff6
> --- /dev/null
> +++ b/scripts/simplebench/simplebench.py
> @@ -0,0 +1,128 @@
> +#!/usr/bin/env python
> +#
> +# Simple benchmarking framework
> +#
> +# Copyright (c) 2019 Virtuozzo International GmbH.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program. If not, see <http://www.gnu.org/licenses/>.
> +#
> +
> +
> +def bench_one(test_func, test_env, test_case, count=5, initial_run=True):
> + """Benchmark one test-case
> +
> + test_func -- benchmarking function with prototype
> + test_func(env, case), which takes test_env and test_case
> + arguments and returns {'seconds': int} (which is benchmark
> + result) on success and {'error': str} on error. Returned
> + dict may contain any other additional fields.
> + test_env -- test environment - opaque first argument for test_func
> + test_case -- test case - opaque second argument for test_func
> + count -- how many times to call test_func, to calculate average
> + initial_run -- do initial run of test_func, which don't get into result
> +
> + Returns dict with the following fields:
> + 'runs': list of test_func results
> + 'average': average seconds per run (exists only if at least one run
> + succeeded)
> + 'delta': maximum delta between test_func result and the average
> + (exists only if at least one run succeeded)
> + 'n-failed': number of failed runs (exists only if at least one run
> + failed)
> + """
> + if initial_run:
> + print(' #initial run:')
> + print(' ', test_func(test_env, test_case))
> +
> + runs = []
> + for i in range(count):
> + print(' #run {}'.format(i+1))
> + res = test_func(test_env, test_case)
> + print(' ', res)
> + runs.append(res)
> +
> + result = {'runs': runs}
> +
> + successed = [r for r in runs if ('seconds' in r)]
> + if successed:
> + avg = sum(r['seconds'] for r in successed) / len(successed)
> + result['average'] = avg
> + result['delta'] = max(abs(r['seconds'] - avg) for r in successed)
> +
> + if len(successed) < count:
> + result['n-failed'] = count - len(successed)
> +
> + return result
> +
> +
> +def ascii_one(result):
> + """Return ASCII representation of bench_one() returned dict."""
> + if 'average' in result:
> + s = '{:.2f} +- {:.2f}'.format(result['average'], result['delta'])
> + if 'n-failed' in result:
> + s += '\n({} failed)'.format(result['n-failed'])
> + return s
> + else:
> + return 'FAILED'
I think it would be visually clearer if "+-" was printed without any
space between it and the following number, using something
like this:
s = ' {:.2f} +-{:.2f}'.format(result['average'], result['delta'])
The resulting table would look like:
---------- ------------- ------------- -------------
backup-1 backup-2 mirror
ssd -> ssd 0.43 +-0.00 4.48 +-0.06 4.38 +-0.02
ssd -> hdd 10.60 +-0.08 10.69 +-0.18 10.57 +-0.05
ssd -> nbd 33.81 +-0.37 10.67 +-0.17 10.07 +-0.07
---------- ------------- ------------- -------------
But, this is just cosmetics.
With or without the suggestion above:
Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
> +
> +
> +def bench(test_func, test_envs, test_cases, *args, **vargs):
> + """Fill benchmark table
> +
> + test_func -- benchmarking function, see bench_one for description
> + test_envs -- list of test environments, see bench_one
> + test_cases -- list of test cases, see bench_one
> + args, vargs -- additional arguments for bench_one
> +
> + Returns dict with the following fields:
> + 'envs': test_envs
> + 'cases': test_cases
> + 'tab': filled 2D array, where cell [i][j] is bench_one result for
> + test_cases[i] for test_envs[j] (i.e., rows are test cases and
> + columns are test environments)
> + """
> + tab = {}
> + results = {
> + 'envs': test_envs,
> + 'cases': test_cases,
> + 'tab': tab
> + }
> + n = 1
> + n_tests = len(test_envs) * len(test_cases)
> + for env in test_envs:
> + for case in test_cases:
> + print('Testing {}/{}: {} :: {}'.format(n, n_tests,
> + env['id'], case['id']))
> + if case['id'] not in tab:
> + tab[case['id']] = {}
> + tab[case['id']][env['id']] = bench_one(test_func, env, case,
> + *args, **vargs)
> + n += 1
> +
> + print('Done')
> + return results
> +
> +
> +def ascii(results):
> + """Return ASCII representation of bench() returned dict."""
> + from tabulate import tabulate
> +
> + tab = [[""] + [c['id'] for c in results['envs']]]
> + for case in results['cases']:
> + row = [case['id']]
> + for env in results['envs']:
> + row.append(ascii_one(results['tab'][case['id']][env['id']]))
> + tab.append(row)
> +
> + return tabulate(tab)
> --
> 2.21.0
>
>
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v3 1/4] scripts/simplebench: add simplebench.py
2020-02-28 13:03 ` Aleksandar Markovic
@ 2020-02-28 13:48 ` Vladimir Sementsov-Ogievskiy
2020-03-02 21:05 ` Aleksandar Markovic
0 siblings, 1 reply; 13+ messages in thread
From: Vladimir Sementsov-Ogievskiy @ 2020-02-28 13:48 UTC (permalink / raw)
To: Aleksandar Markovic
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
28.02.2020 16:03, Aleksandar Markovic wrote:
> On Fri, Feb 28, 2020 at 8:19 AM Vladimir Sementsov-Ogievskiy
> <vsementsov@virtuozzo.com> wrote:
>>
>> Add simple benchmark table creator.
>>
>> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
>> ---
>> scripts/simplebench/simplebench.py | 128 +++++++++++++++++++++++++++++
>> 1 file changed, 128 insertions(+)
>> create mode 100644 scripts/simplebench/simplebench.py
>>
>> diff --git a/scripts/simplebench/simplebench.py b/scripts/simplebench/simplebench.py
>> new file mode 100644
>> index 0000000000..59e7314ff6
>> --- /dev/null
>> +++ b/scripts/simplebench/simplebench.py
>> @@ -0,0 +1,128 @@
>> +#!/usr/bin/env python
>> +#
>> +# Simple benchmarking framework
>> +#
>> +# Copyright (c) 2019 Virtuozzo International GmbH.
>> +#
>> +# This program is free software; you can redistribute it and/or modify
>> +# it under the terms of the GNU General Public License as published by
>> +# the Free Software Foundation; either version 2 of the License, or
>> +# (at your option) any later version.
>> +#
>> +# This program is distributed in the hope that it will be useful,
>> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
>> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>> +# GNU General Public License for more details.
>> +#
>> +# You should have received a copy of the GNU General Public License
>> +# along with this program. If not, see <http://www.gnu.org/licenses/>.
>> +#
>> +
>> +
>> +def bench_one(test_func, test_env, test_case, count=5, initial_run=True):
>> + """Benchmark one test-case
>> +
>> + test_func -- benchmarking function with prototype
>> + test_func(env, case), which takes test_env and test_case
>> + arguments and returns {'seconds': int} (which is benchmark
>> + result) on success and {'error': str} on error. Returned
>> + dict may contain any other additional fields.
>> + test_env -- test environment - opaque first argument for test_func
>> + test_case -- test case - opaque second argument for test_func
>> + count -- how many times to call test_func, to calculate average
>> + initial_run -- do initial run of test_func, which don't get into result
>> +
>> + Returns dict with the following fields:
>> + 'runs': list of test_func results
>> + 'average': average seconds per run (exists only if at least one run
>> + succeeded)
>> + 'delta': maximum delta between test_func result and the average
>> + (exists only if at least one run succeeded)
>> + 'n-failed': number of failed runs (exists only if at least one run
>> + failed)
>> + """
>> + if initial_run:
>> + print(' #initial run:')
>> + print(' ', test_func(test_env, test_case))
>> +
>> + runs = []
>> + for i in range(count):
>> + print(' #run {}'.format(i+1))
>> + res = test_func(test_env, test_case)
>> + print(' ', res)
>> + runs.append(res)
>> +
>> + result = {'runs': runs}
>> +
>> + successed = [r for r in runs if ('seconds' in r)]
>> + if successed:
>> + avg = sum(r['seconds'] for r in successed) / len(successed)
>> + result['average'] = avg
>> + result['delta'] = max(abs(r['seconds'] - avg) for r in successed)
>> +
>> + if len(successed) < count:
>> + result['n-failed'] = count - len(successed)
>> +
>> + return result
>> +
>> +
>> +def ascii_one(result):
>> + """Return ASCII representation of bench_one() returned dict."""
>> + if 'average' in result:
>> + s = '{:.2f} +- {:.2f}'.format(result['average'], result['delta'])
>> + if 'n-failed' in result:
>> + s += '\n({} failed)'.format(result['n-failed'])
>> + return s
>> + else:
>> + return 'FAILED'
>
> I think it would be visually clearer if "+-" was printed without any
> space between it and the following number, using something
> like this:
>
> s = ' {:.2f} +-{:.2f}'.format(result['average'], result['delta'])
>
> The resulting table would look like:
>
> ---------- ------------- ------------- -------------
> backup-1 backup-2 mirror
> ssd -> ssd 0.43 +-0.00 4.48 +-0.06 4.38 +-0.02
> ssd -> hdd 10.60 +-0.08 10.69 +-0.18 10.57 +-0.05
> ssd -> nbd 33.81 +-0.37 10.67 +-0.17 10.07 +-0.07
> ---------- ------------- ------------- -------------
>
> But, this is just cosmetics.
>
> With or without the suggestion above:
>
> Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
Thanks for reviewing!
Agree with this change, but I don't think it worth to resend the series for this one space)
Hope it may be applied with pull request..
>
>> +
>> +
>> +def bench(test_func, test_envs, test_cases, *args, **vargs):
>> + """Fill benchmark table
>> +
>> + test_func -- benchmarking function, see bench_one for description
>> + test_envs -- list of test environments, see bench_one
>> + test_cases -- list of test cases, see bench_one
>> + args, vargs -- additional arguments for bench_one
>> +
>> + Returns dict with the following fields:
>> + 'envs': test_envs
>> + 'cases': test_cases
>> + 'tab': filled 2D array, where cell [i][j] is bench_one result for
>> + test_cases[i] for test_envs[j] (i.e., rows are test cases and
>> + columns are test environments)
>> + """
>> + tab = {}
>> + results = {
>> + 'envs': test_envs,
>> + 'cases': test_cases,
>> + 'tab': tab
>> + }
>> + n = 1
>> + n_tests = len(test_envs) * len(test_cases)
>> + for env in test_envs:
>> + for case in test_cases:
>> + print('Testing {}/{}: {} :: {}'.format(n, n_tests,
>> + env['id'], case['id']))
>> + if case['id'] not in tab:
>> + tab[case['id']] = {}
>> + tab[case['id']][env['id']] = bench_one(test_func, env, case,
>> + *args, **vargs)
>> + n += 1
>> +
>> + print('Done')
>> + return results
>> +
>> +
>> +def ascii(results):
>> + """Return ASCII representation of bench() returned dict."""
>> + from tabulate import tabulate
>> +
>> + tab = [[""] + [c['id'] for c in results['envs']]]
>> + for case in results['cases']:
>> + row = [case['id']]
>> + for env in results['envs']:
>> + row.append(ascii_one(results['tab'][case['id']][env['id']]))
>> + tab.append(row)
>> +
>> + return tabulate(tab)
>> --
>> 2.21.0
>>
>>
--
Best regards,
Vladimir
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v3 1/4] scripts/simplebench: add simplebench.py
2020-02-28 13:48 ` Vladimir Sementsov-Ogievskiy
@ 2020-03-02 21:05 ` Aleksandar Markovic
2020-03-17 14:40 ` Aleksandar Markovic
0 siblings, 1 reply; 13+ messages in thread
From: Aleksandar Markovic @ 2020-03-02 21:05 UTC (permalink / raw)
To: Vladimir Sementsov-Ogievskiy
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
[-- Attachment #1: Type: text/plain, Size: 3758 bytes --]
> >> +
> >> +
> >> +def ascii_one(result):
> >> + """Return ASCII representation of bench_one() returned dict."""
> >> + if 'average' in result:
> >> + s = '{:.2f} +- {:.2f}'.format(result['average'],
result['delta'])
> >> + if 'n-failed' in result:
> >> + s += '\n({} failed)'.format(result['n-failed'])
> >> + return s
> >> + else:
> >> + return 'FAILED'
> >
> > I think it would be visually clearer if "+-" was printed without any
> > space between it and the following number, using something
> > like this:
> >
> > s = ' {:.2f} +-{:.2f}'.format(result['average'], result['delta'])
> >
> > The resulting table would look like:
> >
> > ---------- ------------- ------------- -------------
> > backup-1 backup-2 mirror
> > ssd -> ssd 0.43 +-0.00 4.48 +-0.06 4.38 +-0.02
> > ssd -> hdd 10.60 +-0.08 10.69 +-0.18 10.57 +-0.05
> > ssd -> nbd 33.81 +-0.37 10.67 +-0.17 10.07 +-0.07
> > ---------- ------------- ------------- -------------
> >
> > But, this is just cosmetics.
> >
> > With or without the suggestion above:
> >
> > Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
>
> Thanks for reviewing!
>
> Agree with this change, but I don't think it worth to resend the series
for this one space)
> Hope it may be applied with pull request..
>
I am an occasional Python programmer, and I felt comfortable
reviewing your series, but I am not a maintainer of this directory,
and I believe Eduardo or Cleber or other more active Python
contributors would be better choice for selecting this series in
their pull request.
So, I can't send this series to Peter - Cleber, Eduardo, please
see to it.
Yours,
Aleksandar
> >
> >> +
> >> +
> >> +def bench(test_func, test_envs, test_cases, *args, **vargs):
> >> + """Fill benchmark table
> >> +
> >> + test_func -- benchmarking function, see bench_one for description
> >> + test_envs -- list of test environments, see bench_one
> >> + test_cases -- list of test cases, see bench_one
> >> + args, vargs -- additional arguments for bench_one
> >> +
> >> + Returns dict with the following fields:
> >> + 'envs': test_envs
> >> + 'cases': test_cases
> >> + 'tab': filled 2D array, where cell [i][j] is bench_one
result for
> >> + test_cases[i] for test_envs[j] (i.e., rows are test
cases and
> >> + columns are test environments)
> >> + """
> >> + tab = {}
> >> + results = {
> >> + 'envs': test_envs,
> >> + 'cases': test_cases,
> >> + 'tab': tab
> >> + }
> >> + n = 1
> >> + n_tests = len(test_envs) * len(test_cases)
> >> + for env in test_envs:
> >> + for case in test_cases:
> >> + print('Testing {}/{}: {} :: {}'.format(n, n_tests,
> >> + env['id'],
case['id']))
> >> + if case['id'] not in tab:
> >> + tab[case['id']] = {}
> >> + tab[case['id']][env['id']] = bench_one(test_func, env,
case,
> >> + *args, **vargs)
> >> + n += 1
> >> +
> >> + print('Done')
> >> + return results
> >> +
> >> +
> >> +def ascii(results):
> >> + """Return ASCII representation of bench() returned dict."""
> >> + from tabulate import tabulate
> >> +
> >> + tab = [[""] + [c['id'] for c in results['envs']]]
> >> + for case in results['cases']:
> >> + row = [case['id']]
> >> + for env in results['envs']:
> >> +
row.append(ascii_one(results['tab'][case['id']][env['id']]))
> >> + tab.append(row)
> >> +
> >> + return tabulate(tab)
> >> --
> >> 2.21.0
> >>
> >>
>
>
> --
> Best regards,
> Vladimir
[-- Attachment #2: Type: text/html, Size: 5697 bytes --]
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v3 1/4] scripts/simplebench: add simplebench.py
2020-03-02 21:05 ` Aleksandar Markovic
@ 2020-03-17 14:40 ` Aleksandar Markovic
2020-03-17 15:01 ` Vladimir Sementsov-Ogievskiy
0 siblings, 1 reply; 13+ messages in thread
From: Aleksandar Markovic @ 2020-03-17 14:40 UTC (permalink / raw)
To: Vladimir Sementsov-Ogievskiy
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
On Mon, Mar 2, 2020 at 10:05 PM Aleksandar Markovic
<aleksandar.m.mail@gmail.com> wrote:
>
>
>
>
> > >> +
> > >> +
> > >> +def ascii_one(result):
> > >> + """Return ASCII representation of bench_one() returned dict."""
> > >> + if 'average' in result:
> > >> + s = '{:.2f} +- {:.2f}'.format(result['average'], result['delta'])
> > >> + if 'n-failed' in result:
> > >> + s += '\n({} failed)'.format(result['n-failed'])
> > >> + return s
> > >> + else:
> > >> + return 'FAILED'
> > >
> > > I think it would be visually clearer if "+-" was printed without any
> > > space between it and the following number, using something
> > > like this:
> > >
> > > s = ' {:.2f} +-{:.2f}'.format(result['average'], result['delta'])
> > >
> > > The resulting table would look like:
> > >
> > > ---------- ------------- ------------- -------------
> > > backup-1 backup-2 mirror
> > > ssd -> ssd 0.43 +-0.00 4.48 +-0.06 4.38 +-0.02
> > > ssd -> hdd 10.60 +-0.08 10.69 +-0.18 10.57 +-0.05
> > > ssd -> nbd 33.81 +-0.37 10.67 +-0.17 10.07 +-0.07
> > > ---------- ------------- ------------- -------------
> > >
> > > But, this is just cosmetics.
> > >
> > > With or without the suggestion above:
> > >
> > > Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
> >
> > Thanks for reviewing!
> >
> > Agree with this change, but I don't think it worth to resend the series for this one space)
> > Hope it may be applied with pull request..
> >
>
> I am an occasional Python programmer, and I felt comfortable
> reviewing your series, but I am not a maintainer of this directory,
> and I believe Eduardo or Cleber or other more active Python
> contributors would be better choice for selecting this series in
> their pull request.
>
> So, I can't send this series to Peter - Cleber, Eduardo, please
> see to it.
>
Eduardo, can you perhaps consider this series for selecting
into your pull request?
Thanks,
Aleksandar
> Yours,
> Aleksandar
>
> > >
> > >> +
> > >> +
> > >> +def bench(test_func, test_envs, test_cases, *args, **vargs):
> > >> + """Fill benchmark table
> > >> +
> > >> + test_func -- benchmarking function, see bench_one for description
> > >> + test_envs -- list of test environments, see bench_one
> > >> + test_cases -- list of test cases, see bench_one
> > >> + args, vargs -- additional arguments for bench_one
> > >> +
> > >> + Returns dict with the following fields:
> > >> + 'envs': test_envs
> > >> + 'cases': test_cases
> > >> + 'tab': filled 2D array, where cell [i][j] is bench_one result for
> > >> + test_cases[i] for test_envs[j] (i.e., rows are test cases and
> > >> + columns are test environments)
> > >> + """
> > >> + tab = {}
> > >> + results = {
> > >> + 'envs': test_envs,
> > >> + 'cases': test_cases,
> > >> + 'tab': tab
> > >> + }
> > >> + n = 1
> > >> + n_tests = len(test_envs) * len(test_cases)
> > >> + for env in test_envs:
> > >> + for case in test_cases:
> > >> + print('Testing {}/{}: {} :: {}'.format(n, n_tests,
> > >> + env['id'], case['id']))
> > >> + if case['id'] not in tab:
> > >> + tab[case['id']] = {}
> > >> + tab[case['id']][env['id']] = bench_one(test_func, env, case,
> > >> + *args, **vargs)
> > >> + n += 1
> > >> +
> > >> + print('Done')
> > >> + return results
> > >> +
> > >> +
> > >> +def ascii(results):
> > >> + """Return ASCII representation of bench() returned dict."""
> > >> + from tabulate import tabulate
> > >> +
> > >> + tab = [[""] + [c['id'] for c in results['envs']]]
> > >> + for case in results['cases']:
> > >> + row = [case['id']]
> > >> + for env in results['envs']:
> > >> + row.append(ascii_one(results['tab'][case['id']][env['id']]))
> > >> + tab.append(row)
> > >> +
> > >> + return tabulate(tab)
> > >> --
> > >> 2.21.0
> > >>
> > >>
> >
> >
> > --
> > Best regards,
> > Vladimir
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v3 1/4] scripts/simplebench: add simplebench.py
2020-03-17 14:40 ` Aleksandar Markovic
@ 2020-03-17 15:01 ` Vladimir Sementsov-Ogievskiy
0 siblings, 0 replies; 13+ messages in thread
From: Vladimir Sementsov-Ogievskiy @ 2020-03-17 15:01 UTC (permalink / raw)
To: Aleksandar Markovic
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
17.03.2020 17:40, Aleksandar Markovic wrote:
> On Mon, Mar 2, 2020 at 10:05 PM Aleksandar Markovic
> <aleksandar.m.mail@gmail.com> wrote:
>>
>>
>>
>>
>>>>> +
>>>>> +
>>>>> +def ascii_one(result):
>>>>> + """Return ASCII representation of bench_one() returned dict."""
>>>>> + if 'average' in result:
>>>>> + s = '{:.2f} +- {:.2f}'.format(result['average'], result['delta'])
>>>>> + if 'n-failed' in result:
>>>>> + s += '\n({} failed)'.format(result['n-failed'])
>>>>> + return s
>>>>> + else:
>>>>> + return 'FAILED'
>>>>
>>>> I think it would be visually clearer if "+-" was printed without any
>>>> space between it and the following number, using something
>>>> like this:
>>>>
>>>> s = ' {:.2f} +-{:.2f}'.format(result['average'], result['delta'])
>>>>
>>>> The resulting table would look like:
>>>>
>>>> ---------- ------------- ------------- -------------
>>>> backup-1 backup-2 mirror
>>>> ssd -> ssd 0.43 +-0.00 4.48 +-0.06 4.38 +-0.02
>>>> ssd -> hdd 10.60 +-0.08 10.69 +-0.18 10.57 +-0.05
>>>> ssd -> nbd 33.81 +-0.37 10.67 +-0.17 10.07 +-0.07
>>>> ---------- ------------- ------------- -------------
>>>>
>>>> But, this is just cosmetics.
>>>>
>>>> With or without the suggestion above:
>>>>
>>>> Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
>>>
>>> Thanks for reviewing!
>>>
>>> Agree with this change, but I don't think it worth to resend the series for this one space)
>>> Hope it may be applied with pull request..
>>>
>>
>> I am an occasional Python programmer, and I felt comfortable
>> reviewing your series, but I am not a maintainer of this directory,
>> and I believe Eduardo or Cleber or other more active Python
>> contributors would be better choice for selecting this series in
>> their pull request.
>>
>> So, I can't send this series to Peter - Cleber, Eduardo, please
>> see to it.
>>
>
> Eduardo, can you perhaps consider this series for selecting
> into your pull request?
I saw, he has taken the patches into python-next in:
https://github.com/ehabkost/qemu/commits/python-next
Hope, pull request will come)
>
> Thanks,
> Aleksandar
>
>> Yours,
>> Aleksandar
>>
>>>>
>>>>> +
>>>>> +
>>>>> +def bench(test_func, test_envs, test_cases, *args, **vargs):
>>>>> + """Fill benchmark table
>>>>> +
>>>>> + test_func -- benchmarking function, see bench_one for description
>>>>> + test_envs -- list of test environments, see bench_one
>>>>> + test_cases -- list of test cases, see bench_one
>>>>> + args, vargs -- additional arguments for bench_one
>>>>> +
>>>>> + Returns dict with the following fields:
>>>>> + 'envs': test_envs
>>>>> + 'cases': test_cases
>>>>> + 'tab': filled 2D array, where cell [i][j] is bench_one result for
>>>>> + test_cases[i] for test_envs[j] (i.e., rows are test cases and
>>>>> + columns are test environments)
>>>>> + """
>>>>> + tab = {}
>>>>> + results = {
>>>>> + 'envs': test_envs,
>>>>> + 'cases': test_cases,
>>>>> + 'tab': tab
>>>>> + }
>>>>> + n = 1
>>>>> + n_tests = len(test_envs) * len(test_cases)
>>>>> + for env in test_envs:
>>>>> + for case in test_cases:
>>>>> + print('Testing {}/{}: {} :: {}'.format(n, n_tests,
>>>>> + env['id'], case['id']))
>>>>> + if case['id'] not in tab:
>>>>> + tab[case['id']] = {}
>>>>> + tab[case['id']][env['id']] = bench_one(test_func, env, case,
>>>>> + *args, **vargs)
>>>>> + n += 1
>>>>> +
>>>>> + print('Done')
>>>>> + return results
>>>>> +
>>>>> +
>>>>> +def ascii(results):
>>>>> + """Return ASCII representation of bench() returned dict."""
>>>>> + from tabulate import tabulate
>>>>> +
>>>>> + tab = [[""] + [c['id'] for c in results['envs']]]
>>>>> + for case in results['cases']:
>>>>> + row = [case['id']]
>>>>> + for env in results['envs']:
>>>>> + row.append(ascii_one(results['tab'][case['id']][env['id']]))
>>>>> + tab.append(row)
>>>>> +
>>>>> + return tabulate(tab)
>>>>> --
>>>>> 2.21.0
>>>>>
>>>>>
>>>
>>>
>>> --
>>> Best regards,
>>> Vladimir
--
Best regards,
Vladimir
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v3 2/4] scripts/simplebench: add qemu/bench_block_job.py
2020-02-28 7:19 [PATCH v3 0/4] benchmark util Vladimir Sementsov-Ogievskiy
2020-02-28 7:19 ` [PATCH v3 1/4] scripts/simplebench: add simplebench.py Vladimir Sementsov-Ogievskiy
@ 2020-02-28 7:19 ` Vladimir Sementsov-Ogievskiy
2020-02-28 13:04 ` Aleksandar Markovic
2020-02-28 7:19 ` [PATCH v3 3/4] scripts/simplebench: add example usage of simplebench Vladimir Sementsov-Ogievskiy
2020-02-28 7:19 ` [PATCH v3 4/4] MAINTAINERS: add simplebench Vladimir Sementsov-Ogievskiy
3 siblings, 1 reply; 13+ messages in thread
From: Vladimir Sementsov-Ogievskiy @ 2020-02-28 7:19 UTC (permalink / raw)
To: qemu-devel
Cc: kwolf, vsementsov, ehabkost, qemu-block, stefanha, mreitz, crosa,
den, jsnow
Add block-job benchmarking helper functions.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
---
scripts/simplebench/bench_block_job.py | 119 +++++++++++++++++++++++++
1 file changed, 119 insertions(+)
create mode 100755 scripts/simplebench/bench_block_job.py
diff --git a/scripts/simplebench/bench_block_job.py b/scripts/simplebench/bench_block_job.py
new file mode 100755
index 0000000000..9808d696cf
--- /dev/null
+++ b/scripts/simplebench/bench_block_job.py
@@ -0,0 +1,119 @@
+#!/usr/bin/env python
+#
+# Benchmark block jobs
+#
+# Copyright (c) 2019 Virtuozzo International GmbH.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+
+import sys
+import os
+import socket
+import json
+
+sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', 'python'))
+from qemu.machine import QEMUMachine
+from qemu.qmp import QMPConnectError
+
+
+def bench_block_job(cmd, cmd_args, qemu_args):
+ """Benchmark block-job
+
+ cmd -- qmp command to run block-job (like blockdev-backup)
+ cmd_args -- dict of qmp command arguments
+ qemu_args -- list of Qemu command line arguments, including path to Qemu
+ binary
+
+ Returns {'seconds': int} on success and {'error': str} on failure, dict may
+ contain addional 'vm-log' field. Return value is compatible with
+ simplebench lib.
+ """
+
+ vm = QEMUMachine(qemu_args[0], args=qemu_args[1:])
+
+ try:
+ vm.launch()
+ except OSError as e:
+ return {'error': 'popen failed: ' + str(e)}
+ except (QMPConnectError, socket.timeout):
+ return {'error': 'qemu failed: ' + str(vm.get_log())}
+
+ try:
+ res = vm.qmp(cmd, **cmd_args)
+ if res != {'return': {}}:
+ vm.shutdown()
+ return {'error': '"{}" command failed: {}'.format(cmd, str(res))}
+
+ e = vm.event_wait('JOB_STATUS_CHANGE')
+ assert e['data']['status'] == 'created'
+ start_ms = e['timestamp']['seconds'] * 1000000 + \
+ e['timestamp']['microseconds']
+
+ e = vm.events_wait((('BLOCK_JOB_READY', None),
+ ('BLOCK_JOB_COMPLETED', None),
+ ('BLOCK_JOB_FAILED', None)), timeout=True)
+ if e['event'] not in ('BLOCK_JOB_READY', 'BLOCK_JOB_COMPLETED'):
+ vm.shutdown()
+ return {'error': 'block-job failed: ' + str(e),
+ 'vm-log': vm.get_log()}
+ end_ms = e['timestamp']['seconds'] * 1000000 + \
+ e['timestamp']['microseconds']
+ finally:
+ vm.shutdown()
+
+ return {'seconds': (end_ms - start_ms) / 1000000.0}
+
+
+# Bench backup or mirror
+def bench_block_copy(qemu_binary, cmd, source, target):
+ """Helper to run bench_block_job() for mirror or backup"""
+ assert cmd in ('blockdev-backup', 'blockdev-mirror')
+
+ source['node-name'] = 'source'
+ target['node-name'] = 'target'
+
+ return bench_block_job(cmd,
+ {'job-id': 'job0', 'device': 'source',
+ 'target': 'target', 'sync': 'full'},
+ [qemu_binary,
+ '-blockdev', json.dumps(source),
+ '-blockdev', json.dumps(target)])
+
+
+def drv_file(filename):
+ return {'driver': 'file', 'filename': filename,
+ 'cache': {'direct': True}, 'aio': 'native'}
+
+
+def drv_nbd(host, port):
+ return {'driver': 'nbd',
+ 'server': {'type': 'inet', 'host': host, 'port': port}}
+
+
+if __name__ == '__main__':
+ import sys
+
+ if len(sys.argv) < 4:
+ print('USAGE: {} <qmp block-job command name> '
+ '<json string of arguments for the command> '
+ '<qemu binary path and arguments>'.format(sys.argv[0]))
+ exit(1)
+
+ res = bench_block_job(sys.argv[1], json.loads(sys.argv[2]), sys.argv[3:])
+ if 'seconds' in res:
+ print('{:.2f}'.format(res['seconds']))
+ else:
+ print(res)
--
2.21.0
^ permalink raw reply related [flat|nested] 13+ messages in thread
* Re: [PATCH v3 2/4] scripts/simplebench: add qemu/bench_block_job.py
2020-02-28 7:19 ` [PATCH v3 2/4] scripts/simplebench: add qemu/bench_block_job.py Vladimir Sementsov-Ogievskiy
@ 2020-02-28 13:04 ` Aleksandar Markovic
0 siblings, 0 replies; 13+ messages in thread
From: Aleksandar Markovic @ 2020-02-28 13:04 UTC (permalink / raw)
To: Vladimir Sementsov-Ogievskiy
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
On Fri, Feb 28, 2020 at 8:19 AM Vladimir Sementsov-Ogievskiy
<vsementsov@virtuozzo.com> wrote:
>
> Add block-job benchmarking helper functions.
>
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> ---
Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
> scripts/simplebench/bench_block_job.py | 119 +++++++++++++++++++++++++
> 1 file changed, 119 insertions(+)
> create mode 100755 scripts/simplebench/bench_block_job.py
>
> diff --git a/scripts/simplebench/bench_block_job.py b/scripts/simplebench/bench_block_job.py
> new file mode 100755
> index 0000000000..9808d696cf
> --- /dev/null
> +++ b/scripts/simplebench/bench_block_job.py
> @@ -0,0 +1,119 @@
> +#!/usr/bin/env python
> +#
> +# Benchmark block jobs
> +#
> +# Copyright (c) 2019 Virtuozzo International GmbH.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program. If not, see <http://www.gnu.org/licenses/>.
> +#
> +
> +
> +import sys
> +import os
> +import socket
> +import json
> +
> +sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', 'python'))
> +from qemu.machine import QEMUMachine
> +from qemu.qmp import QMPConnectError
> +
> +
> +def bench_block_job(cmd, cmd_args, qemu_args):
> + """Benchmark block-job
> +
> + cmd -- qmp command to run block-job (like blockdev-backup)
> + cmd_args -- dict of qmp command arguments
> + qemu_args -- list of Qemu command line arguments, including path to Qemu
> + binary
> +
> + Returns {'seconds': int} on success and {'error': str} on failure, dict may
> + contain addional 'vm-log' field. Return value is compatible with
> + simplebench lib.
> + """
> +
> + vm = QEMUMachine(qemu_args[0], args=qemu_args[1:])
> +
> + try:
> + vm.launch()
> + except OSError as e:
> + return {'error': 'popen failed: ' + str(e)}
> + except (QMPConnectError, socket.timeout):
> + return {'error': 'qemu failed: ' + str(vm.get_log())}
> +
> + try:
> + res = vm.qmp(cmd, **cmd_args)
> + if res != {'return': {}}:
> + vm.shutdown()
> + return {'error': '"{}" command failed: {}'.format(cmd, str(res))}
> +
> + e = vm.event_wait('JOB_STATUS_CHANGE')
> + assert e['data']['status'] == 'created'
> + start_ms = e['timestamp']['seconds'] * 1000000 + \
> + e['timestamp']['microseconds']
> +
> + e = vm.events_wait((('BLOCK_JOB_READY', None),
> + ('BLOCK_JOB_COMPLETED', None),
> + ('BLOCK_JOB_FAILED', None)), timeout=True)
> + if e['event'] not in ('BLOCK_JOB_READY', 'BLOCK_JOB_COMPLETED'):
> + vm.shutdown()
> + return {'error': 'block-job failed: ' + str(e),
> + 'vm-log': vm.get_log()}
> + end_ms = e['timestamp']['seconds'] * 1000000 + \
> + e['timestamp']['microseconds']
> + finally:
> + vm.shutdown()
> +
> + return {'seconds': (end_ms - start_ms) / 1000000.0}
> +
> +
> +# Bench backup or mirror
> +def bench_block_copy(qemu_binary, cmd, source, target):
> + """Helper to run bench_block_job() for mirror or backup"""
> + assert cmd in ('blockdev-backup', 'blockdev-mirror')
> +
> + source['node-name'] = 'source'
> + target['node-name'] = 'target'
> +
> + return bench_block_job(cmd,
> + {'job-id': 'job0', 'device': 'source',
> + 'target': 'target', 'sync': 'full'},
> + [qemu_binary,
> + '-blockdev', json.dumps(source),
> + '-blockdev', json.dumps(target)])
> +
> +
> +def drv_file(filename):
> + return {'driver': 'file', 'filename': filename,
> + 'cache': {'direct': True}, 'aio': 'native'}
> +
> +
> +def drv_nbd(host, port):
> + return {'driver': 'nbd',
> + 'server': {'type': 'inet', 'host': host, 'port': port}}
> +
> +
> +if __name__ == '__main__':
> + import sys
> +
> + if len(sys.argv) < 4:
> + print('USAGE: {} <qmp block-job command name> '
> + '<json string of arguments for the command> '
> + '<qemu binary path and arguments>'.format(sys.argv[0]))
> + exit(1)
> +
> + res = bench_block_job(sys.argv[1], json.loads(sys.argv[2]), sys.argv[3:])
> + if 'seconds' in res:
> + print('{:.2f}'.format(res['seconds']))
> + else:
> + print(res)
> --
> 2.21.0
>
>
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v3 3/4] scripts/simplebench: add example usage of simplebench
2020-02-28 7:19 [PATCH v3 0/4] benchmark util Vladimir Sementsov-Ogievskiy
2020-02-28 7:19 ` [PATCH v3 1/4] scripts/simplebench: add simplebench.py Vladimir Sementsov-Ogievskiy
2020-02-28 7:19 ` [PATCH v3 2/4] scripts/simplebench: add qemu/bench_block_job.py Vladimir Sementsov-Ogievskiy
@ 2020-02-28 7:19 ` Vladimir Sementsov-Ogievskiy
2020-02-28 13:27 ` Aleksandar Markovic
2020-02-28 7:19 ` [PATCH v3 4/4] MAINTAINERS: add simplebench Vladimir Sementsov-Ogievskiy
3 siblings, 1 reply; 13+ messages in thread
From: Vladimir Sementsov-Ogievskiy @ 2020-02-28 7:19 UTC (permalink / raw)
To: qemu-devel
Cc: kwolf, vsementsov, ehabkost, qemu-block, stefanha, mreitz, crosa,
den, jsnow
This example may be used as a template for custom benchmark.
It illustrates three things to prepare:
- define bench_func
- define test environments (columns)
- define test cases (rows)
And final call of simplebench API.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
---
scripts/simplebench/bench-example.py | 80 ++++++++++++++++++++++++++++
1 file changed, 80 insertions(+)
create mode 100644 scripts/simplebench/bench-example.py
diff --git a/scripts/simplebench/bench-example.py b/scripts/simplebench/bench-example.py
new file mode 100644
index 0000000000..c642a5b891
--- /dev/null
+++ b/scripts/simplebench/bench-example.py
@@ -0,0 +1,80 @@
+#!/usr/bin/env python3
+#
+# Benchmark example
+#
+# Copyright (c) 2019 Virtuozzo International GmbH.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+
+import simplebench
+from bench_block_job import bench_block_copy, drv_file, drv_nbd
+
+
+def bench_func(env, case):
+ """ Handle one "cell" of benchmarking table. """
+ return bench_block_copy(env['qemu_binary'], env['cmd'],
+ case['source'], case['target'])
+
+
+# You may set the following five variables to correct values, to turn this
+# example to real benchmark.
+ssd_source = '/path-to-raw-source-image-at-ssd'
+ssd_target = '/path-to-raw-target-image-at-ssd'
+hdd_target = '/path-to-raw-source-image-at-hdd'
+nbd_ip = 'nbd-ip-addr'
+nbd_port = 'nbd-port-number'
+
+# Test-cases are "rows" in benchmark resulting table, 'id' is a caption for
+# the row, other fields are handled by bench_func.
+test_cases = [
+ {
+ 'id': 'ssd -> ssd',
+ 'source': drv_file(ssd_source),
+ 'target': drv_file(ssd_target)
+ },
+ {
+ 'id': 'ssd -> hdd',
+ 'source': drv_file(ssd_source),
+ 'target': drv_file(hdd_target)
+ },
+ {
+ 'id': 'ssd -> nbd',
+ 'source': drv_file(ssd_source),
+ 'target': drv_nbd(nbd_ip, nbd_port)
+ },
+]
+
+# Test-envs are "columns" in benchmark resulting table, 'id is a caption for
+# the column, other fields are handled by bench_func.
+test_envs = [
+ {
+ 'id': 'backup-1',
+ 'cmd': 'blockdev-backup',
+ 'qemu_binary': '/path-to-qemu-binary-1'
+ },
+ {
+ 'id': 'backup-2',
+ 'cmd': 'blockdev-backup',
+ 'qemu_binary': '/path-to-qemu-binary-2'
+ },
+ {
+ 'id': 'mirror',
+ 'cmd': 'blockdev-mirror',
+ 'qemu_binary': '/path-to-qemu-binary-1'
+ }
+]
+
+result = simplebench.bench(bench_func, test_envs, test_cases, count=3)
+print(simplebench.ascii(result))
--
2.21.0
^ permalink raw reply related [flat|nested] 13+ messages in thread
* Re: [PATCH v3 3/4] scripts/simplebench: add example usage of simplebench
2020-02-28 7:19 ` [PATCH v3 3/4] scripts/simplebench: add example usage of simplebench Vladimir Sementsov-Ogievskiy
@ 2020-02-28 13:27 ` Aleksandar Markovic
0 siblings, 0 replies; 13+ messages in thread
From: Aleksandar Markovic @ 2020-02-28 13:27 UTC (permalink / raw)
To: Vladimir Sementsov-Ogievskiy
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
On Fri, Feb 28, 2020 at 8:21 AM Vladimir Sementsov-Ogievskiy
<vsementsov@virtuozzo.com> wrote:
>
> This example may be used as a template for custom benchmark.
> It illustrates three things to prepare:
> - define bench_func
> - define test environments (columns)
> - define test cases (rows)
> And final call of simplebench API.
>
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> ---
In future, it would be perhaps useful to add option to output results
in the format of csv (comma separated values) (or similar) - it would
simplify import to other data presentation tools. In any case:
Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
> scripts/simplebench/bench-example.py | 80 ++++++++++++++++++++++++++++
> 1 file changed, 80 insertions(+)
> create mode 100644 scripts/simplebench/bench-example.py
>
> diff --git a/scripts/simplebench/bench-example.py b/scripts/simplebench/bench-example.py
> new file mode 100644
> index 0000000000..c642a5b891
> --- /dev/null
> +++ b/scripts/simplebench/bench-example.py
> @@ -0,0 +1,80 @@
> +#!/usr/bin/env python3
> +#
> +# Benchmark example
> +#
> +# Copyright (c) 2019 Virtuozzo International GmbH.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program. If not, see <http://www.gnu.org/licenses/>.
> +#
> +
> +import simplebench
> +from bench_block_job import bench_block_copy, drv_file, drv_nbd
> +
> +
> +def bench_func(env, case):
> + """ Handle one "cell" of benchmarking table. """
> + return bench_block_copy(env['qemu_binary'], env['cmd'],
> + case['source'], case['target'])
> +
> +
> +# You may set the following five variables to correct values, to turn this
> +# example to real benchmark.
> +ssd_source = '/path-to-raw-source-image-at-ssd'
> +ssd_target = '/path-to-raw-target-image-at-ssd'
> +hdd_target = '/path-to-raw-source-image-at-hdd'
> +nbd_ip = 'nbd-ip-addr'
> +nbd_port = 'nbd-port-number'
> +
> +# Test-cases are "rows" in benchmark resulting table, 'id' is a caption for
> +# the row, other fields are handled by bench_func.
> +test_cases = [
> + {
> + 'id': 'ssd -> ssd',
> + 'source': drv_file(ssd_source),
> + 'target': drv_file(ssd_target)
> + },
> + {
> + 'id': 'ssd -> hdd',
> + 'source': drv_file(ssd_source),
> + 'target': drv_file(hdd_target)
> + },
> + {
> + 'id': 'ssd -> nbd',
> + 'source': drv_file(ssd_source),
> + 'target': drv_nbd(nbd_ip, nbd_port)
> + },
> +]
> +
> +# Test-envs are "columns" in benchmark resulting table, 'id is a caption for
> +# the column, other fields are handled by bench_func.
> +test_envs = [
> + {
> + 'id': 'backup-1',
> + 'cmd': 'blockdev-backup',
> + 'qemu_binary': '/path-to-qemu-binary-1'
> + },
> + {
> + 'id': 'backup-2',
> + 'cmd': 'blockdev-backup',
> + 'qemu_binary': '/path-to-qemu-binary-2'
> + },
> + {
> + 'id': 'mirror',
> + 'cmd': 'blockdev-mirror',
> + 'qemu_binary': '/path-to-qemu-binary-1'
> + }
> +]
> +
> +result = simplebench.bench(bench_func, test_envs, test_cases, count=3)
> +print(simplebench.ascii(result))
> --
> 2.21.0
>
>
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v3 4/4] MAINTAINERS: add simplebench
2020-02-28 7:19 [PATCH v3 0/4] benchmark util Vladimir Sementsov-Ogievskiy
` (2 preceding siblings ...)
2020-02-28 7:19 ` [PATCH v3 3/4] scripts/simplebench: add example usage of simplebench Vladimir Sementsov-Ogievskiy
@ 2020-02-28 7:19 ` Vladimir Sementsov-Ogievskiy
2020-02-28 13:27 ` Aleksandar Markovic
3 siblings, 1 reply; 13+ messages in thread
From: Vladimir Sementsov-Ogievskiy @ 2020-02-28 7:19 UTC (permalink / raw)
To: qemu-devel
Cc: kwolf, vsementsov, ehabkost, qemu-block, stefanha, mreitz, crosa,
den, jsnow
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
---
MAINTAINERS | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 5e5e3e52d6..16d069adc5 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2038,6 +2038,11 @@ F: python/qemu/*py
F: scripts/*.py
F: tests/*.py
+Benchmark util
+M: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
+S: Maintained
+F: scripts/simplebench/
+
QAPI
M: Markus Armbruster <armbru@redhat.com>
M: Michael Roth <mdroth@linux.vnet.ibm.com>
--
2.21.0
^ permalink raw reply related [flat|nested] 13+ messages in thread
* Re: [PATCH v3 4/4] MAINTAINERS: add simplebench
2020-02-28 7:19 ` [PATCH v3 4/4] MAINTAINERS: add simplebench Vladimir Sementsov-Ogievskiy
@ 2020-02-28 13:27 ` Aleksandar Markovic
0 siblings, 0 replies; 13+ messages in thread
From: Aleksandar Markovic @ 2020-02-28 13:27 UTC (permalink / raw)
To: Vladimir Sementsov-Ogievskiy
Cc: Kevin Wolf, Eduardo Habkost, open list:bochs, Stefan Hajnoczi,
QEMU Developers, Max Reitz, Cleber Rosa, Denis V. Lunev,
John Snow
On Fri, Feb 28, 2020 at 8:19 AM Vladimir Sementsov-Ogievskiy
<vsementsov@virtuozzo.com> wrote:
>
> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> ---
Reviewed-by: Aleksandar Markovic <amarkovic@wavecomp.com>
> MAINTAINERS | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 5e5e3e52d6..16d069adc5 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -2038,6 +2038,11 @@ F: python/qemu/*py
> F: scripts/*.py
> F: tests/*.py
>
> +Benchmark util
> +M: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> +S: Maintained
> +F: scripts/simplebench/
> +
> QAPI
> M: Markus Armbruster <armbru@redhat.com>
> M: Michael Roth <mdroth@linux.vnet.ibm.com>
> --
> 2.21.0
>
>
^ permalink raw reply [flat|nested] 13+ messages in thread