Hi Bart,
I'm sorry for delayed response. It's examination time here at my university
and today it's going to end :)
On Tue, Apr 19, 2011 at 5:19 PM, Bart van Andel wrote:
> Hi Dip,
>
> I'm sorry but I fail to see the importance of these preliminary results.
>
> You did not provide detailed enough info about the used equipment to see
> what the results mean. My guess is that the Atom equipped computer contains
> far less memory than the Core one, and obviously it has less processing
> power.
>
>
I'm sorry again that i didn't uploaded all details in full. Actually i'm
doing some initial testing here to understand different scenarios for
performance test. During this i think i gave less attention towards wiki
documentation. Anyways i'll fix it with few more interesting test results
soon.
For the time being please find configurations of machine used here :
machine-1: http://goo.gl/XRQcY and machine-2: http://goo.gl/MO2hp
Also raw data for all these tests here: http://goo.gl/GXVm4
> Some observations:
> - The CPU usage graphs look pretty similar when stretched to 100% of the
> total time used (so each device takes 100%) instead of time in seconds.
> - This is also valid for the memory write graph (what is read/secs? KB/s?
> Read operations per second?)
> - About the same is true for the memory graphs, although of course the Core
> uses more memory than the Atom. Is this total amount of memory used by the
> whole system or just by Hugin? I'm asking because (max mem) - (min mem) is
> about 200MB for both devices. And please use MB instead of kB on the
> vertical axis, this makes it more readable.
>
This initial testing was for overall system when all other activities on
system were nominal. (Next step would be to test only on
Hugin+it's dependencies processes only). Sure i'll take care of using units
which will little easy to read hence forth.
> - Because the Atom has less memory *and* less speed it can't read all the
> images at once but spreads it out over more time than the Core. This is all
> pretty logical and I don't see what information your tests have provided
> that we couldn't know by simple reasoning.
>
>
In my test, Core was around 1.83125 times faster than atom. If core took 51
secs to complete the given task then atom should finish it in 51*1.83125 =
~93.4 secs when rest of the system and software configurations are made
same. But atom took ~124 secs to complete the task. Question is if i take
small single board computer like this http://goo.gl/CdQ9j will hugin run on
it? How long does it take to perform the task etc?
> Furthermore, the notion that
>
> Calculating/finding the
> bottlenecks and tuning expensive algorithms to boost this performance
> should be valued contribution to hugin community.
>
> always makes sense, it does not need a speed test. Tuning algorithms
> (especially memory intensive ones, where not all data will fit in RAM) for
> optimal performance and memory usage is a nice aim, but I don't think that's
> what you propose to do yourself. At least your proposal doesn't mention it.
>
>
Yes, I must have to measure the performance of atleast core algorithms
(probably on different hardware).
> I've taken a look at the schedule in your proposal too and don't get me
> wrong, but you're taking a lot of time to create a working XML
> parser/builder. I hope you're aware that there are libraries available which
> will do this task for you without you having to worry about implementation
> details, right? For instance, PHP has SimpleXML and also XmlReader/XmlWriter
> which are very easy to use and definitely don't need 2 months work.
>
>
When i said to build XML core engine meaning to build a system that will
listen for responses for various scatter hugin clients and will take actions
accordingly. I have to intension to build yet another xml reader/writer or
parser etc. Probably some miss-communication happed from mine side!
> Last point: why would I want a tab to evaluate Hugin speed? I can see a
> point in a menu item for this, but something as permanently visible as a tab
> does not make sense to me.
>
> Overall I think your proposal definitely needs some work. I hope you didn't
> think this reply is too discouraging, I'm just trying to see the actual
> point in this.
>
>
Discouraging? Nope, not at all! I love this part. Different thoughts, views
are always welcome. This helps a lot while understanding the problem
statements at greater extent...
Thanks.
Best,
Dip
> --
> Bart
>
> --
> You received this message because you are subscribed to the Google Groups
> "Hugin and other free panoramic software" group.
> A list of frequently asked questions is available at:
> http://wiki.panotools.org/Hugin_FAQ
> To post to this group, send email to hugin-ptx@googlegroups.com
> To unsubscribe from this group, send email to
> hugin-ptx+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/hugin-ptx
>
--
You received this message b
