Re: [PATCH v3 1/4] scripts/simplebench: add simplebench.py

[Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>]

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