Constant keying

classic Classic list List threaded Threaded
16 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Constant keying

Fever Havannah

Hi guys,

For the past day or so svxlink has been keying my radio every 20 to 30 seconds non-stop ...

Any pointers?

Danny
ZS6XOX


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

ka2pbt
could you be in some "oscillation cycle" .. the noise from the squelch close triggers the squelch open .... happens when I turn on Parrot through a repeater .... 

--> Rob, KA2PBT
721st Mechanized Contest Battalion



On Tue, Dec 1, 2015 at 6:05 PM, Fever Havannah <[hidden email]> wrote:

Hi guys,

For the past day or so svxlink has been keying my radio every 20 to 30 seconds non-stop ...

Any pointers?

Danny
ZS6XOX


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel



------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Fever Havannah

There you go ... squelch was on the edge ... I've set it up a little higher ... seems to be o.k now ...

Thanks Rob

Danny
ZS6XOX
73's

On 01 Dec 2015 8:19 PM, "ka2pbt" <[hidden email]> wrote:
could you be in some "oscillation cycle" .. the noise from the squelch close triggers the squelch open .... happens when I turn on Parrot through a repeater .... 

--> Rob, KA2PBT
721st Mechanized Contest Battalion



On Tue, Dec 1, 2015 at 6:05 PM, Fever Havannah <[hidden email]> wrote:

Hi guys,

For the past day or so svxlink has been keying my radio every 20 to 30 seconds non-stop ...

Any pointers?

Danny
ZS6XOX


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel



------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

David Ranch-3

Set up a PL tone on your radio (if it supports COS) or set it up Svxlink
itself.  That will stop Svxlink from thinking there is a valid signal
and thus keying up your radio to give a courtesy beep.

--David
KI6ZHD


On 12/01/2015 10:22 AM, Fever Havannah wrote:

>
> There you go ... squelch was on the edge ... I've set it up a little
> higher ... seems to be o.k now ...
>
> Thanks Rob
>
> Danny
> ZS6XOX
> 73's
>


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Rob Janssen
David Ranch wrote:
> Set up a PL tone on your radio (if it supports COS) or set it up Svxlink
> itself.  That will stop Svxlink from thinking there is a valid signal
> and thus keying up your radio to give a courtesy beep.
>
> --David
> KI6ZHD
>

Of course it is important to also solve the root cause, e.g. using PL/CTCSS and by setting
the squelch correctly, but we also implemented a small delay for opening the repeater on
squelch:

OPEN_ON_SQL=250

and to suppress the courtesy beep when very short transmissions (kerchunkers) are made:

#
# Keep the start of transmission timestamp to detect short transmissions
#
variable start_of_transmission 0;
variable length_of_transmission 0;

#
# Executed each time the squelch is opened or closed
#   rx_id   - The ID of the RX that the squelch opened/closed on
#   is_open - Set to 1 if the squelch is open or 0 if it's closed
#
proc squelch_open {rx_id is_open} {
   variable sql_rx_id;
   variable start_of_transmission;
   variable length_of_transmission;

   #puts "The squelch is $is_open on RX $rx_id";

   if {$is_open} {
     set start_of_transmission [clock clicks -milliseconds];
     set length_of_transmission 0;
   } else {
     if { $start_of_transmission > 0 } {
       set length_of_transmission [expr [clock clicks -milliseconds] - $start_of_transmission];
     }
     if { $length_of_transmission > 1500 } {
       set sql_rx_id $rx_id;
     }
     set start_of_transmission 0;
   }
}

The result is that short squelch openings do not result in continuous beeping.

By the way, when you have unfiltered audio out of the receiver it is often better to use the
svxlink siglev mechanism (with SQL_DET=SIGLEV and SIGLEV_DET=NOISE plus proper
calibration of the SIGLEV mechanism) instead of the receiver hardware squelch.
It works well, and it has independent settings for OPEN and CLOSE so a suitable hysteresis
can be configured to avoid spurious openings while still having a sensitive receiver.

Rob

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

David Ranch-3

Hello Rob,

> OPEN_ON_SQL=250

Oh.. that's a pretty nice feature that I wasn't aware of!


> and to suppress the courtesy beep when very short transmissions (kerchunkers) are made:

If the above timer is in place, why is the below code change needed?


--David

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Fever Havannah

Thanks Rob ...

On 02 Dec 2015 6:12 PM, "David Ranch" <[hidden email]> wrote:

Hello Rob,

> OPEN_ON_SQL=250

Oh.. that's a pretty nice feature that I wasn't aware of!


> and to suppress the courtesy beep when very short transmissions (kerchunkers) are made:

If the above timer is in place, why is the below code change needed?


--David

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Rob Janssen
In reply to this post by David Ranch-3
David Ranch wrote:
>> and to suppress the courtesy beep when very short transmissions (kerchunkers) are made:
> If the above timer is in place, why is the below code change needed?
>

They are two different things.  The first item is to prevent the repeater to key up on short interference bursts
or very short transmissions.   The second item is to suppress the courtesy tone, once the repeater is open,
on short transmissions.  That threshold is somewhat higher on our repeater (1.5 seconds).  So you need to
make a longish transmission to get back a courtesy tone, but you can key the repeater with a 0.25 second
transmission.  That way the kerchunkers that just want to check if they open the squelch do not trigger the
courtesy tone.

BTW, that routine also requires an "if { $sql_rx_id > 0} {" to be added around the code in send_rgr_sound,
something we already had for other reasons.

Rob

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

David Ranch-3

Hello Rob,

>
> They are two different things.  The first item is to prevent the repeater to key up on short interference bursts
> or very short transmissions.   The second item is to suppress the courtesy tone, once the repeater is open,
> on short transmissions.

Ok, that makes sense now.  Your patch seems to me seems pretty valuable
so maybe we can get it patched into the official sources?


> BTW, that routine also requires an "if { $sql_rx_id > 0} {" to be added around the code in send_rgr_sound,
> something we already had for other reasons.

Could you post your complete change as a proper "diff -u" format?

--David

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Rob Janssen
David Ranch wrote:
> Hello Rob,
>
>> They are two different things.  The first item is to prevent the repeater to key up on short interference bursts
>> or very short transmissions.   The second item is to suppress the courtesy tone, once the repeater is open,
>> on short transmissions.
> Ok, that makes sense now.  Your patch seems to me seems pretty valuable
> so maybe we can get it patched into the official sources?

We'll have to ask Tobias' opinion...
>
>
>> BTW, that routine also requires an "if { $sql_rx_id > 0} {" to be added around the code in send_rgr_sound,
>> something we already had for other reasons.
> Could you post your complete change as a proper "diff -u" format?

Of course we have many other changes in send_rgr_sound, but when you start from the standard version in Logic.tcl it
would end up like this:

#
# Executed when the squelch just have closed and the RGR_SOUND_DELAY timer has
# expired.
#
proc send_rgr_sound {} {
   variable sql_rx_id;

   if { $sql_rx_id > 0} {
     playTone 440 500 100;
     playSilence 200;

     for {set i 0} {$i < $sql_rx_id} {incr i 1} {
       playTone 880 500 50;
       playSilence 50;
     }
     playSilence 100;
     set sql_rx_id 0;
   }
}

(a diff would show many changes as the indent level has been increased)

Rob

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Tobias Blomberg
On Wednesday 02 December 2015 19:14:38 Rob Janssen wrote:

> David Ranch wrote:
> > Hello Rob,
> >
> >> They are two different things.  The first item is to prevent the repeater
> >> to key up on short interference bursts or very short transmissions.  
> >> The second item is to suppress the courtesy tone, once the repeater is
> >> open, on short transmissions.
> >
> > Ok, that makes sense now.  Your patch seems to me seems pretty valuable
> > so maybe we can get it patched into the official sources?
>
> We'll have to ask Tobias' opinion...

A feature like that is better implemented in the SvxLink C++ core. I'm sure it
works just fine in TCL as well but I want to keep the TCL event handlers as
free from implementing features as possible. It's nice that things like that
can be done in TCL for specific installations but for a thing like that to be
included in the SvxLink distribution it would have to be coded in C++. It's a
nice feature and probably not that hard to implement.

What would we call the config variable?  KERCHUNK_RGR_SOUND_ELIM :-)

73's de SM0SVX / Tobias


>
> >> BTW, that routine also requires an "if { $sql_rx_id > 0} {" to be added
> >> around the code in send_rgr_sound, something we already had for other
> >> reasons.
> >
> > Could you post your complete change as a proper "diff -u" format?
>
> Of course we have many other changes in send_rgr_sound, but when you start
> from the standard version in Logic.tcl it would end up like this:
>
> #
> # Executed when the squelch just have closed and the RGR_SOUND_DELAY timer
> has # expired.
> #
> proc send_rgr_sound {} {
>    variable sql_rx_id;
>
>    if { $sql_rx_id > 0} {
>      playTone 440 500 100;
>      playSilence 200;
>
>      for {set i 0} {$i < $sql_rx_id} {incr i 1} {
>        playTone 880 500 50;
>        playSilence 50;
>      }
>      playSilence 100;
>      set sql_rx_id 0;
>    }
> }
>
> (a diff would show many changes as the indent level has been increased)
>
> Rob


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Constant keying

Rob Janssen
SM0SVX wrote:

> On Wednesday 02 December 2015 19:14:38 Rob Janssen wrote:
>> David Ranch wrote:
>>> Hello Rob,
>>>
>>>> They are two different things.  The first item is to prevent the repeater
>>>> to key up on short interference bursts or very short transmissions.
>>>> The second item is to suppress the courtesy tone, once the repeater is
>>>> open, on short transmissions.
>>> Ok, that makes sense now.  Your patch seems to me seems pretty valuable
>>> so maybe we can get it patched into the official sources?
>> We'll have to ask Tobias' opinion...
> A feature like that is better implemented in the SvxLink C++ core. I'm sure it
> works just fine in TCL as well but I want to keep the TCL event handlers as
> free from implementing features as possible. It's nice that things like that
> can be done in TCL for specific installations but for a thing like that to be
> included in the SvxLink distribution it would have to be coded in C++. It's a
> nice feature and probably not that hard to implement.
>
> What would we call the config variable?  KERCHUNK_RGR_SOUND_ELIM :-)
>
> 73's de SM0SVX / Tobias
>

I know that is your position on issues like this...
Of course it means that every envisionable functionality would have to be coded and configurable.
The TCL solution adds some flexibility for local wishes.

For example, what is not shown in that code is that we also have a proc:

# Function to clear sql_rx_id to be called when echolink squelch closes
# Added to suppress the normal rgr_sound sent above

proc clear_sql_rx_id {} {
   variable sql_rx_id;

   set sql_rx_id 0;
}

which is called from proc is_receiving in the EchoLink.tcl.  The result is that echolink transmissions
have only the courtesy beep defined in that proc in the echolink module, not followed by the normal
courtesy beep defined in Logic.tcl.  Without that, users are confused if a transmission is from
EchoLink or from radio, and also there was the issue that the radio beep reflected the receiver used
for the last radio transmission, which is not related to the EchoLink transmission at all.
(remember we have many receivers!)

A change like that would probably be considered "too dirty" to include in the EchoLink C++ code,
but I think in the TCL code it is accepable (I deliberately used a proc to do the resetting, instead of
trying to access the global variable directly)

Also, in our snd_rgr_sound routine we check for long transmissions as well:

   set rnd [randmax 14];

   if { $sql_rx_id > 0 && $length_of_transmission > 250000 && $rnd < 10 } {
     playSilence 150;
     playMsg "transmissiontime" $rnd;
     playSilence 150;
   }

where randmax is a proc:

# generate random integer number in the range [0,max]

proc randmax {max} {
   return [expr {int(rand()*($max+1))}]
}

What it does: when somebody talks for more than 4 minutes, a random message from the transmissiontime
directory, where we have 10 messages named 0.wav .. 9.wav, is played to tell the speaker in different ways
to talk a bit shorter.   Just a gimmick, but it kind of works.   It is probably not a good plan to code all things
like that in the C++ core, because for new users the number of options would just be overwhelming.

Rob

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Multithreading

Mathias Koch DL9BQ
Hi,

is it possible, that SvxLink only uses one CPU core?
In a htop I miss the multithreading.
I have a RTL-SDR installed here. Works great. Fantastic sound.
But in my Banana PI is the second CPU at 80%, and the first only 10%.
Unfortunately the SvxLink process crashes always after a very short time.

Best regards
73 Mathias


------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Multithreading

Rob Janssen
Mathias Koch DL9BQ wrote:

> Hi,
>
> is it possible, that SvxLink only uses one CPU core?
> In a htop I miss the multithreading.
> I have a RTL-SDR installed here. Works great. Fantastic sound.
> But in my Banana PI is the second CPU at 80%, and the first only 10%.
> Unfortunately the SvxLink process crashes always after a very short time.
>
> Best regards
> 73 Mathias
>

That is right, the main svxlink task is single threaded so it can use only a single CPU.
There is a thread for communication with the RTL stick, but it is not the place where the SDR code is running
so it uses almost no CPU.

Indeed, this means that a multicore system will not be used in the most efficient way.
However, this should not be related to crashing.  When you have crashes, there is probably another problem.
Do you get an error message when it crashes?  Try to run it interactively from the prompt to get more chance
of seeing the message about the crash.

To offload part of the code to another thread you can also use a remotetrx on the same system, where you
have the RTL-SDR receiver(s) configured, and connect it from the main svxlink process via IP address 127.0.0.1.
Then at least you are using two cores.

Rob

------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Multithreading

Mathias Koch DL9BQ
Hi Rob,

thanks for the information.
That's too bad!
I had already the idea with the remotetrx  ...
But not great ...
The crashes are on the system load.
This can for example be provoke even with the parrot.

73 Mathias

RJ> Mathias Koch DL9BQ wrote:

>> Hi,
>>
>> is it possible, that SvxLink only uses one CPU core?
>> In a htop I miss the multithreading.
>> I have a RTL-SDR installed here. Works great. Fantastic sound.
>> But in my Banana PI is the second CPU at 80%, and the first only 10%.
>> Unfortunately the SvxLink process crashes always after a very short time.
>>
>> Best regards
>> 73 Mathias
>>

RJ> That is right, the main svxlink task is single threaded so it can use only a single CPU.
RJ> There is a thread for communication with the RTL stick, but it is
RJ> not the place where the SDR code is running
RJ> so it uses almost no CPU.

RJ> Indeed, this means that a multicore system will not be used in the most efficient way.
RJ> However, this should not be related to crashing.  When you have
RJ> crashes, there is probably another problem.
RJ> Do you get an error message when it crashes?  Try to run it
RJ> interactively from the prompt to get more chance
RJ> of seeing the message about the crash.

RJ> To offload part of the code to another thread you can also use a
RJ> remotetrx on the same system, where you
RJ> have the RTL-SDR receiver(s) configured, and connect it from the
RJ> main svxlink process via IP address 127.0.0.1.
RJ> Then at least you are using two cores.

RJ> Rob



------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741911&iu=/4140
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Multithreading

Tobias Blomberg
On Sunday 06 December 2015 19:20:16 Mathias Koch DL9BQ wrote:
> Hi Rob,
>
> thanks for the information.
> That's too bad!

Yes, the implementation of the SDR stuff has not been fully optimized yet. It
works just fine but threading is something I want to implement to be able to
run SvxLink with SDR on the RPi for example.


> I had already the idea with the remotetrx  ...
> But not great ...

This is the only work-around I can think of at the moment. I did it this way
when testing SvxLink with RTL-SDR on RPi 2 and it seemed to work quite well.
But of course, threading the SDR code would be much better.

73's de SM0SVX / Tobias


> The crashes are on the system load.
> This can for example be provoke even with the parrot.
>
> 73 Mathias
>
> RJ> Mathias Koch DL9BQ wrote:
> >> Hi,
> >>
> >> is it possible, that SvxLink only uses one CPU core?
> >> In a htop I miss the multithreading.
> >> I have a RTL-SDR installed here. Works great. Fantastic sound.
> >> But in my Banana PI is the second CPU at 80%, and the first only 10%.
> >> Unfortunately the SvxLink process crashes always after a very short time.
> >>
> >> Best regards
> >> 73 Mathias
>
> RJ> That is right, the main svxlink task is single threaded so it can use
> only a single CPU. RJ> There is a thread for communication with the RTL
> stick, but it is RJ> not the place where the SDR code is running
> RJ> so it uses almost no CPU.
>
> RJ> Indeed, this means that a multicore system will not be used in the most
> efficient way. RJ> However, this should not be related to crashing.  When
> you have RJ> crashes, there is probably another problem.
> RJ> Do you get an error message when it crashes?  Try to run it
> RJ> interactively from the prompt to get more chance
> RJ> of seeing the message about the crash.
>
> RJ> To offload part of the code to another thread you can also use a
> RJ> remotetrx on the same system, where you
> RJ> have the RTL-SDR receiver(s) configured, and connect it from the
> RJ> main svxlink process via IP address 127.0.0.1.
> RJ> Then at least you are using two cores.
>
> RJ> Rob


------------------------------------------------------------------------------
_______________________________________________
Svxlink-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/svxlink-devel
Loading...