[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: Resource and energy efficiency of KDE applications
From:       David Hurka <david.hurka () mailbox ! org>
Date:       2021-06-29 23:24:54
Message-ID: 5624667.MhkbZ0Pkbq () doro
[Download RAW message or body]

On Wednesday, June 30, 2021 12:57:48 AM CEST Cornelius Schumacher wrote:
> On 29.06.21 19:38, Arjun AK wrote:
> > I don't think measuring current on a Desktop is the right way to do
> > this. Your desktop is going to be too "noisy" when it comes to power
> > measurements. Activity from components like hard disk, WiFi/BT, Ethernet
> > or other software during the testing would end up affecting the
> > measurements.

Like noticed in the measurements at Umweltcampus, this noise can be filtered 
out by running multiple tests.

> > It would be better off doing this on a SBC (maybe a PI2?),
> > with as much hardware turned off as possible and running a minimum set
> > of software. Measurements could be done with a multimeter[1] or maybe
> > even a INA219[2].
> 
> Part of the measurement is to also record system utilisation such as
> harddisk or network activity. That makes it possible to at least to some
> degree account for the noisy part.

Indeed. And on a typical desktop computer we still have the opportunity to 
measure power consumption of individual components.

> It's also a tradeoff between measuring in a realistic scenario which is
> close to normal use and a prepared setup which is aimed at getting more
> exact numbers.
> 
> I think it would be interesting to do both and see what the numbers say.

I think it would also make sense to compare direct power measurements to 
software metrics, like tick counters or the power display in KSysguard.

> There certainly is a lot to be learned and to be improved here. For me
> the most important step is to be more aware of energy consumption and to
> try to get numbers at all. Any insight is welcome :-)

+1 :)

Cheers, David



[prev in list] [next in list] [prev in thread] [next in thread]