This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Add support for multiple data points to perf tests
- From: Yao Qi <qiyaoltc at gmail dot com>
- To: Doug Evans <xdje42 at gmail dot com>
- Cc: Yao Qi <qiyaoltc at gmail dot com>, "gdb-patches\ at sourceware dot org" <gdb-patches at sourceware dot org>
- Date: Wed, 29 Jul 2015 09:45:08 +0100
- Subject: Re: [PATCH] Add support for multiple data points to perf tests
- Authentication-results: sourceware.org; auth=none
- References: <m3k2tmzwxh dot fsf at seba dot sebabeach dot org> <86pp3dfzsd dot fsf at gmail dot com> <CAP9bCMSg0L_kXp80uP0qjKaRiXxTxU73nQtESOObm7yvGP9Asw at mail dot gmail dot com>
Doug Evans <xdje42@gmail.com> writes:
> All I've done is rename perftest.log to perftest.sum,
> and then put more test results in text form in perftest.log.
Oh, I see.
> I *could* put it all in one file, but I like what's in perftest.log today,
> (though there are a few tweaks I'd like to make),
> I'm just moving it to perftest.sum.
> Following the pattern of file naming already in use
> will be helpful to everyone using the perf testsuite.
Right. To be clear, with your patch applied, raw data is saved in
perftest.sum and some basic statistical analysis is output to
perftest.log.
>
> If/when we have other forms of output I would expect them
> to not go to .log or .sum files.
>
I agree.
>>> I want to record the raw data somewhere, and a bit of statistical analysis
>>> (standard deviation left for another day), but I also don't want
>>> it to clutter up the basic report.
>>> This patch takes a cue from gdb.{sum,log} and does the same thing
>>> with perftest.{sum,log}.
>>
>> I'd like pertest framework only writes raw data down in some formats
>> (such as text, xml or jason) without any analysis. Statistical analysis
>> can be performed in a post-mortem way by other scripts.
>
> We can generate a .csv file with the raw data if/when the need arises.
> [or whatever is easiest for post-processing scripts to read]
>
> A human readable form with basic statistical analysis
> is useful today, and doable in 5 lines of code.
> Let's add the complexity when there's a demonstrated
> need for it.
OK, no problem. I don't have other questions, and your patch looks good
to me.
--
Yao (éå)