[SLL] technically speaking, what might the vendor do?

[Derek Simkowiak]

/> ...made no difference in the camera's performance./

    Dang.  I thought that would be it.

    You mentioned it was a C++ app (w/SWIG to talk to Java).  It might 
be useful to get a profile of your code, so you can see which function 
call is taking so much time (and yet running smoothly as root).  The 
particular function name might give a data point as to what the actual 
bottleneck is.

    The GNU profiler can be used for this.  A decent intro article on it 
can be found here:

http://www.ibm.com/developerworks/library/l-gnuprof.html

    Besides looking at the source code, doing some more general system 
profiling may be necessary to get some actual data on where the 
bottleneck is.  Here's one tool of many:

http://www.daimi.au.dk/~sandmann/sysprof/ 
<http://www.daimi.au.dk/%7Esandmann/sysprof/>

    Another step, if you have time, would be to follow Glenn's 
suggestion (2008-12-02 11:20AM) and install the real time kernel (just 
as a test if nothing else). 


    Presently we don't know where the bottleneck is at.  Possibly it's 
the scheduler (because root processes do get higher priority than 
non-root), possibly the GigE network connection (although I don't know 
why root would be fast here, but non-root would be slow), or perhaps 
something in your non-root environment (like an LD_LIBRARY_PATH setting 
is loading a bloated debug library or something weird like that).  If we 
can reduce the bottleneck from 'something on the computer' to a specific 
hardware device, function call, or kernel feature, we could narrow down 
the list of things to try...


--Derek

Mathew D. Watson wrote:
> I was all excited when I saw Rob's posting. ... Unfortunately, it 
> didn't make a difference. The default max frequency on my system was 
> 64. Increasing it to 1024, or reducing it to 30, made no difference in 
> the camera's performance.
>
> Thank you very much for the suggestion. I appreciate it.
>
> Mat
>
>
>
>> Did you try changing the RTC max user frequency values?
>>
>>    Nice call!  My money is on Rob's RTC frequency fix...
>>
>> --Derek
>>
>> Rob Smith wrote:
>>> On Tue, Dec 2, 2008 at 9:10 AM, Mathew D. Watson
>>> <watson at visiongate3d.com> wrote:
>>>  
>>>> There is a large performance difference between running the camera 
>>>> app as
>>>> root and running the app at 30 fsp and as a regular user, at a -20 
>>>> nice
>>>> value, at 10 fps. The fraction of corrupted images drops by 
>>>> dramatically (a
>>>> factor of a 1000 IIRC) when running as root.
>>>>     
>>>
>>> Did you try changing the RTC max user frequency values?
>>>
>>> http://www.mythtv.org/wiki/index.php/Frame_display_timing#Adjusting_the_RTC_Interrupt_Frequency 
>>>
>>>
>>> That's a fairly big difference for users vs root out of the box.
>>>   
>>
>>
>>