RE: [ECOS] Performance timing

[C B <csb_80 at hotmail dot com>]

System profiling is probably what I need.

Unfortunately there isn't an implementation of the profiling timer for my target.  Any information or guidance available that you're aware of for implementing this?  For what it's worth, my target is a arm based SAM9 variant.

Thanks.


> Date: Fri, 16 Nov 2007 11:14:51 -0700
> From: gary@mlbassoc.com
> To: csb_80@hotmail.com
> CC: ecos-discuss@ecos.sourceware.org
> Subject: Re: [ECOS] Performance timing
>
> C B wrote:
>>
>> I'm interested in collecting some performance measures on some of my own functions. Are there some simple utilities or example for doing this or is looking at tm_basic.cxx the best bet?
>
> It depends on the scope of what you want to measure. The
> techniques used by tm_basic are good for small things - those
> that execute in less than a system tick (typically 10ms), but
> using them requires some invasive changes (you have to explicitly
> capture start and stop times for a function/segment, etc).
>
> Alternatively, you can turn on system profiling. All this
> requires is a [fairly] high resolution timer and some RAM.
> There is no need to modify your code directly. The data
> is gathered into a RAM buffer and then offloaded (the only
> method for this currently uses the network and TFTP). Once
> you have the data off the box, it can be analyzed using
> standard tools, such as 'gprof'
>
> --
> ------------------------------------------------------------
> Gary Thomas | Consulting for the
> MLB Associates | Embedded world
> ------------------------------------------------------------

_________________________________________________________________
Windows Live Hotmail and Microsoft Office Outlook – together at last. ?Get it now.
http://office.microsoft.com/en-us/outlook/HA102225181033.aspx?pid=CL100626971033

--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss