discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Flow Graph Reconfiguration


From: Gould, Daniel N.
Subject: Re: [Discuss-gnuradio] Flow Graph Reconfiguration
Date: Fri, 5 Sep 2014 11:24:12 +0000

The source is the omsosdr-rtl block and the filter is the FIR. The blocks in 
the modem are OOT modules. Both flow graphs work independently, but when I try 
to dynamically switch between the modems, I get the same error you are seeing. 
I do have a buffer requirement change on the output of the FIR when I switch 
modems, but the querying the FIR object tells me that the min_output_buffer is 
correct.

Thanks

-----Original Message-----
From: Kevin Reid [mailto:address@hidden On Behalf Of Kevin Reid
Sent: Thursday, September 04, 2014 10:58 PM
To: Gould, Daniel N.
Cc: address@hidden
Subject: Re: [Discuss-gnuradio] Flow Graph Reconfiguration

On Sep 4, 2014, at 13:27, Gould, Daniel N. <address@hidden> wrote:

> FlowGraph 1:
> [src] -> [filt] -> [modem-1]
>  
> FlowGraph 2:
> [src] -> [filt] -> [modem-2]
>  
> FG-1 runs to completion, is then disconnected. Then FG-2 is connected and 
> executed. For both flowgraphs the src and filt are the same. The modems are 
> switched out and as such the requirements on the filter output buffer size 
> changes. The filter min_output_buffer size is adjusted following the 
> disconnect from FG-1 and prior to connecting FG-2.
>  
> When FG-2 is executed I immediately get an error from the scheduler 
> complaining that a downstream sample requirement can't be met.

What actual block classes are you using?

I've recently found a bug with the same effect as you describe, which occurs 
when I reuse a pfb_arb_resampler_ccf (but not involving any explicit buffer 
control). See my report: http://gnuradio.org/redmine/issues/713

A workaround I found was to redesign the graph so that I could use a 
rational_resampler instead, but I don't know if this bug is truly specific to 
that block or if the change avoids the problem by accident.

-- 
Kevin Reid                                  <http://switchb.org/kpreid/>




reply via email to

[Prev in Thread] Current Thread [Next in Thread]