V6-pre observations

[john....@gmail.com]

Hello Leif,

Thank you for V6.00-pre.

Starting a new thread for V6 reports.

r1072, xlinrad64, cw/ssb

Hardware:  Airspy HF+ and Asus Xonar U5 soundcard for audio out.

If FFT2/hires window is enabled, after the RX is started, if you click to change RX frequency on baseband or wideband graph, the screen freezes.  

If FFT2 is disabled, RX seems to work OK.  

Windows version:

There seems to be an issue with audio output.  Get an error "INTERNAL ERROR: 1017 [1017] routine:open_audio_output  file:lsetad.c

Could not open device for audio output.
Use U = A/D and D/A setup in main menu to select proper drivers."

The happens even if the output soundcard is disabled.

Please let me know if you need additional information.

73,

John N2NC

[SP2BPD]

I'm moving here my message from the other thread:

Hi Leif and All,

I've tried to start Linrad rev. 1072 with soundcard (actually with 3 different soundcard plus Afedri USB) with no success.

Always ended with:

"Linrad RX input from: PORTAUDIO device number ## : INVALID DEVICE".

Tested on my 2 computers, both Windows 10.

Afedri-NET is accepted on input but cannot assign output device via Portaudio:

"Linrad RX output to: PORTAUDIO device number ## : INVALID DEVICE"

Selecting output soundcard without engaging Portaudio works only halfway. I have motion picture in Baseband window but no sound in a speaker. 

Kind regards

Piotr, SP2BPD

[john....@gmail.com]

Clarification on FFT2 enabled issue.  Only the baseband waterfall and spectrum freezes.  Wideband and hires window remain functional.

-John N2NC

[john....@gmail.com]

Hello Leif,

Some more reports on V6 (r1078)

xlinrad64 and windows (32bit linrad.exe on 64 bit windows10 PC):

The horizontal size of the baseband window can generate errors (stack smashing, segmentation fault or illegal instruction on Linux).  I ran into this while resizing the baseband window.  Resizing the window horizontal size can also make the baseband spectrum disappear at certain sizes.

Windows (32 bit, built on 64 bit Windows 10).

If using portaudio with linrad.exe, if you try and change parameter "Output sampling speed (Hz)", the program crashes.  I tried all available versions of portaudio.dll

Let me know if you require more details.  Input in all cases is Airspy HF+

Thank You,

John N2NC

[Leif Asbrink]

Hello John,

When you resize the baseband window horizontal size Linrad may
change the fft size and rearrange memory allocations. Do you
see the baseband spectrum disappear permanently at certain sizes?

What mode are you in? SSB, AM,... and what processing parameters
do you have in the baseband window? Memory allocations may differ.

Have you tried with portaudio with 64 bit linrad.exe?

Portaudio typically gives a choice between several sound systems.
MME, direct sound, wasapi etc. Do they all crasch if you change
the output speed?

73

Leif

> --
> There is an excellent Linrad User Guide by Gaetan, ON4KHG, at:
> http://w3sz.com/Linrad%20Installation%20&%20Configuration%20User%20Guide%20-%20V1-0.pdf
> ---
> You received this message because you are subscribed to the Google Groups "Linrad" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to linrad+un...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/linrad/eece5212-f4f3-4c4e-b1aa-51cb4b6f92e8n%40googlegroups.com.

[john....@gmail.com]

Hello Leif,

I will research more and report the details here.  This was all CW.  

The spectrum disappearing seemed to be a specific horizontal window size.    Crashes were more likely than spectrum disappearing.

I typically run about 48 kHz sample rate into the baseband and FFT3 bandwidth is typically between 50-100 Hz.  This gives me low latency for a real time receiver.

I have not tried 64 bit linrad.exe.  Will try that.

Portaudio was WDM-KS.  I will try additional sound systems.

73,

John n2nc

[john....@gmail.com]

Hello Leif,

I will put the windows portaudio and "disappearing baseband spectrum" aside for now.

Here is more information on the crashes that depend on baseband window size.  I can recreate the problem on linrad.exe, linrad64.exe and xlinrad64.exe.  Unfortunately, the baseband window sizes were different across all 3 cases.  I could recreate the problems on 2 different PCs for all 3 executables.

Note that the crash does not require the receiver to start,  As soon as I press "B" to start CW, there is a crash if baseband horizontal size is wrong.  I need to manually edit par_cw_bg to stop the crash.

I used my phone's high speed video recording to record the screen before crash.  linrad.exe and linrad64.exe do not finish "drawing" the baseband window before the crash, linrad64.exe seems to build the full baseband window before crash.

I placed a summary of my findings and parameter files on my google drive.  

https://drive.google.com/drive/folders/1Q-rGqna3Zw4kq6CL8RTxhTGOcoZ4i6__?usp=drive_link

The spreadsheet here lists the baseband window xleft and xright settings that causes crashes.

https://docs.google.com/spreadsheets/d/1iAX7qu1_nhlzcKyPn1ECXiZ5f-8oNABQZo0SVMXq5Fo/edit?usp=sharing

I hope this is useful information.

73,

John N2NC

[john....@gmail.com]

Correction:

 linrad.exe and xlinrad64.exe do not finish "drawing" the baseband window before the crash, linrad64.exe seems to build the full baseband window before crash.

[Leif Asbrink]

Hello John,

I made a fresh download of 1078 and changed the serial nubmers for airspy HF.
Then i started xlinrad64 and there was no crash
The par_cw_bg parameters were 485 and 1565 and there should be no crash
according toyour spreadsheet.

I changed to 486 and 1565. Still no problem. I have resized the baseband
by moving around both the left and the right hand borders with and without
having the rx running.

This is the system:
root@debian:/home/bsz/dsp# lscpu
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 39 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Vendor ID: GenuineIntel
BIOS Vendor ID: Intel(R) Corporation
Model name: Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz
BIOS Model name: Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz To Be Filled
By O.E.M. CPU @ 3.5GHz
BIOS CPU family: 207
CPU family: 6
Model: 158
Thread(s) per core: 2
Core(s) per socket: 8
Socket(s): 1
Stepping: 12
CPU(s) scaling MHz: 22%
CPU max MHz: 5000.0000
CPU min MHz: 800.0000
BogoMIPS: 7200.00

I will try some other machines tomorrow.

73

Leif

> To view this discussion visit https://groups.google.com/d/msgid/linrad/7a23541c-5bde-45cb-9d25-22b088bdaa1dn%40googlegroups.com.

[Leif Asbrink]

Hi again John,

Try to set a somewhat smaller screen width. You have set 1950
while my screen is 1920. This seems to work OK, but I have to move the
xlinrad window to see what happens at the edges.

The screen size may cause the problem. There are several arrays
that have the screen width as their size. Try something like 1800
to see if that changes how crashes occur.

73

Leif

[john....@gmail.com]

Hello Leif,

I tried making the window smaller and I still notice the same behavior.  I added some new entries to my spreadsheet with screen width at 1300.

https://docs.google.com/spreadsheets/d/1iAX7qu1_nhlzcKyPn1ECXiZ5f-8oNABQZo0SVMXq5Fo/edit?usp=sharing

I my have found a new clue.  If the baseband window width is between 936-943 pixels, the error "SCREEN OVERLOAD" is displayed on the wide graph.  At 944 pixels it crashes.

lscpu dump from the two systems I am using is below signature.  Not as powerful as your system.

73,

- John N2NC

n2nc@I5-HP-N2NC:~/linrad-code-r1078-trunk$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 36 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 4 On-line CPU(s) list: 0-3 Vendor ID: GenuineIntel Model name: Intel(R) Core(TM) i5-3475S CPU @ 2.90GHz CPU family: 6 Model: 58 Thread(s) per core: 1 Core(s) per socket: 4 Socket(s): 1 Stepping: 9 CPU(s) scaling MHz: 54% CPU max MHz: 3600.0000 CPU min MHz: 1600.0000 BogoMIPS: 5786.45

n2nc@n2nc-NUC6i5SYK:~$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 39 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 4 On-line CPU(s) list: 0-3 Vendor ID: GenuineIntel Model name: Intel(R) Core(TM) i5-6260U CPU @ 1.80GHz CPU family: 6 Model: 78 Thread(s) per core: 2 Core(s) per socket: 2 Socket(s): 1 Stepping: 3 CPU(s) scaling MHz: 47% CPU max MHz: 2900.0000 CPU min MHz: 400.0000 BogoMIPS: 3600.00

[Leif Asbrink]

Hello John,

I was able to get the error on an elderly computer running Windows 7.
It took quite some time for me to find the error. I had written
zero to a string to make sure it would fit on the available
space, but forgot to check that I did not write outside the allocated
space. That means a zero byte was placed on the stack at a position
depending on the width of the basebad window with unpredictable
consequences.

r1079 is now available on the repo.

73

Leif

[john....@gmail.com]

Thank you Leif.  I will try it later today.

73,

John

[john....@gmail.com]

r1079 working good here Leif -- thank you.

- John N2NC