[Linrad] Re: Some succes on an embedded ALIX system

[Xtof <gmail.com; xtof.huygens@xxxxxxxxxxxxxxxx>]

Hi Leif et al,

I turned of the eyecandy and gained around 5%. Except for the
decimation on the SDR-IQ which has a big impact, the others seem to be
less important, gaining only a few %. Also, changing the color depth
only gave me a few %.

I now reliably run a 110Khz slice in CW, with power consumption of
around 10W, for the embedded card-HD combo....nice. I do get choppy
audio once in a while, but toggling (not changing for 2>1, just
toggling) seems to solve this (for say 10 minutes or so).
Good enough for my microwave purposes.

I ll do some more structured changes soon. In top, linrad takes about
55%, and Xorg about 15% - that s it!

73 Xtof on4iy.

On 5 jan, 22:06, Leif Asbrink <l...@xxxxxxxxxx> wrote:
> Hello Xtof,
>
> > I did get a question I forgot with a range 10-999Hz to answer at
> > hardware setup,
> > but I could not find what this about so started with 100.
>
> That was "max DMA rate"
>
> Linrad will set the buffer size for input and output
> devices and through that it forces the hardware to
> its direct memory access (DMA) at a high rate in case
> the user enters bandwidths and other parameters that
> could produce a short dealy from antenna to loudspeaker.
>
> On a slow computer one might want to limit the DMA rate
> to avoid the extra overhead.
>
> > On normal CW, this runs fine, with about 60% cpu. No choppy audio
> > unless one starts other stuff.
> > On weak signal CW, audio is choppy and CPU at 130%. The rest functions
> > fine.
>
> > I ll see if I can optimize to squeeze out a little more. Any hints
> > appreciated, I do need X since
> > the system also runs my HPIB controlled band switching GUI.
>
> There are many things you can do to make Linrad run more
> efficiently. Start with X11 itself.
>
> 1) Make sure you do not have "visual effects" of any
>    kind enabled. (Ubuntu, at least recent versions have them
>    by default and they destroy the performance of X11)
>
> 2) Try to set a smaller colour depth. Linrad is happy with
>    a colour depth of 256 only so if your other softwares
>    can accept that it would save some resources.
>
> Besides setting the SDR-IQ decimation factors high enough
> to make the input data stream slow enough you could
> compromise with the window functions. In case you do
> not need a high dynamic range you might go for a sine
> (to power 1) window. Check the different fft versions.
> on some CPUs they might differ significantly.
>
> I guess the problem with weak signal CW is the second
> fft. In case you need the noise blanker, it should run
> fine in case the CPU has MMX instructions so you can
> select those fft2 variants.
>
> A large "first mixer bandwidth reduction" as well as a low
> sampling rate for the loudspeaker output should also
> save some resources.
>
> Please report further progress:-)
>
> 73
>
> Leif / SM5BSZ
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Linrad" group.
To post to this group, send email to linrad@xxxxxxxxxxxxxxxx
To unsubscribe from this group, send email to linrad+unsubscribe@xxxxxxxxxxxxxxxx
For more options, visit this group at http://groups.google.com/group/linrad?hl=en
-~----------~----~----~----~------~----~------~--~---


LINRADDARNIL