I had a test session with one user who can easily reproduce the audio dropouts.
Latency testing shows his system more than capable of handling realtime
audio.I got some timings from the dataSink.The Good plot is a normal/good
decode. One high time period is corrected on the next iteration.The plots show
the time intervals between calls to dataSink.
The Bad plot is when decoding has stopped. the n_ihsym never gets to 50 to
satisfy the n_ihsym requirement
if(m_ihsym == m_hsymStop) {
Starting another program (most any one) will cause decoding to stop. Having
another audio program start will cause decoding to start succeeding again. So
the audio gets stalled periodically and sticks there until something tickles it.
de Mike W9MDB
On Saturday, January 12, 2019, 12:03:15 AM CST, Mike Lewis
<k7...@hotmail.com> wrote:
Running the modern weather app set to radar observation map, it plays an
animated loop and decodes work fine then, just the occasional cycle missed.
This should have no audio connection, just cpu. CPU is hanging around 15% with
that running (and meinberg and n1mm).
The SSD is SK Hynix BC501 HFM256GDJTNG-8310A - a 256GB stick drive. PCIe Gen
3x2. Likely M.2 form factor.
https://www.skhynix.com/eng/product/ssdClient.jsp scroll down for BC501. Not
too shabby perf.
Don’t forget there are 2 separate sound devices this happens on, the USB and
the internal Realtek. Happens with either input. Can have 2 instances of WSJT
running on both inputs, and JTDX and all 3 instances stop and start at the same
time. They all share the same WSJTX library though. We know the sound is
getting through since the wave files and any audio app like Audacity keep
recording audio fine. Has to be timing or maybe high noise I am thinking.
Unfortunately to get low noise I have to go up to 28 and there are no signals
there these days, save the occasional one off I made contact with today.
From: Black Michael <mdblac...@yahoo.com>
Sent: Saturday, January 12, 2019 00:50
To: Mike Lewis <k7...@hotmail.com>
Subject: Re: [wsjt-devel] Odd Decoding Problem on new Core i5-8250 Laptop -
Only decodes perfect when running HDSDR in parallel
Uninstall and reinstall the sound device.
What kind of SSD do you have? There were some problems before with SSDs but
not about sound as far as I know.
On Friday, January 11, 2019, 11:38:27 PM CST, Mike Lewis <k7...@hotmail.com>
wrote:
I have changed affinity, priority. Audacity when recording has the same
healing effect as HDSDR. I turned off the firewall, no change. I just turned
on programs performance vs auto, no decodes when N1mm is running still. I
tried playing with N1MM recording test button but no real difference either.
Timing could certainly be a factor. Testing the high background noise theory I
tried all manner of settings in the K3 for Noise Blanker (DSP and/or IF) and
Noise reduction, APF, and manual notch, in combination with different filter
settings and RF gain and level. No change, maybe a few chance decodes on a
transition of major adjustments.
Do not have many programs on the machine that do something, but after changing
program perf, I started Meinberg monitor and sat on the status page. It is
decoding a portion of the stations on some cycles, pretty spotty still, but
better. Turn N1MM off, all decode good. Turn it back on, barely any decode
cycles, so back to square one.
So for all my theories, I am no further. My second antenna is a 6M dipole in
the attic and is just as noisy.
From: Black Michael <mdblac...@yahoo.com>
Sent: Saturday, January 12, 2019 00:21
To: Mike Lewis <k7...@hotmail.com>
Subject: Re: [wsjt-devel] Odd Decoding Problem on new Core i5-8250 Laptop -
Only decodes perfect when running HDSDR in parallel
Check your Performance settings and ensure it's set for "Programs".
And perhaps increase the priority for WSJT-X and see if any of those changes
affect it.
Starting another app would take some CPU.
I'm doing some analysis on the timing now to see if that reveals anything.
Nothing obvious yet other than it looks like the sound samples are getting
starved.
On Friday, January 11, 2019, 9:58:21 PM CST, Mike Lewis <k7...@hotmail.com>
wrote:
I started browsing the code on sourceforge tracking down m_ihsym. I see
symspec is called which I found in lib\symspec.f90. Line 87 is where it gets
incremented. It is in index to the number of half symbols. My interpretation
is that the decode period is allowed to run until enough half symbol time slots
have passed. For FT8 the max, m_hymStop, is set to 50.
It would seem something happens near the end of the period that cause the data
block getting built to exit out before hitting completion at 50. Could be a
timer read issue, a file error, or, thinking about how noisy my Camano location
is, perhaps the quality checks in symspec are looking at excessive noise and
bailing out. I see the computation is followed by quality checks that involve
spectra power that appear to be drastic if I read them right (and I might
certainly not be reading them right!).
If I stop HDSDR halfway into a FT8 cycle, I get 4 out the 9 stations decoded in
that period on the old laptop. Playing that wav file on my desktop I only get
the same 3 decoded stations. So whatever changes on the start and stop of
HDSDR or N1MM is captured in the wave file and any computer will fail to decode
the other 6 stations.
- I think the new laptop is making some sort of noise energy we cannot
readily see that affect the quality and that is captured in the wav file. See
the attached wav files for the 4 of 9 stations decoded on the new laptop and
the other file is the same audio going to the old laptop where it decoded 9 of
9.
To get a longer wav file I switched to WSPR mode and spun the dial to an FT8
frequency and let it record. WSPR prints time stamps for each 2 minute period
completed. I stopped HDSDR and like in FT8, it stopped decoding also.
- I believe this problem affects most or all modes, at least the ones that
use this same library code for quality.
- Mike
From: Black Michael <mdblac...@yahoo.com>
Sent: Friday, January 11, 2019 17:32
To: Mike Lewis <k7...@hotmail.com>
Subject: Re: [wsjt-devel] Odd Decoding Problem on new Core i5-8250 Laptop -
Only decodes perfect when running HDSDR in parallel
Can I hook up with you?
Call me 321-690-2551
Mike
On Friday, January 11, 2019, 3:38:07 PM CST, Mike Lewis <k7...@hotmail.com>
wrote:
I do not see it create the log file. It is creating a lock file in
appdata\temp, and a wsjt folder in temp, but it only has a lock, decoded, and
status file in it. Did a disk search, no joy.
From: Black Michael <mdblac...@yahoo.com>
Sent: Friday, January 11, 2019 16:20
To: Mike Lewis <k7...@hotmail.com>
Subject: Re: [wsjt-devel] Odd Decoding Problem on new Core i5-8250 Laptop -
Only decodes perfect when running HDSDR in parallel
OK...here's a new build....this an installation package so should run fine.
And it should create that same file that I need.
https://www.dropbox.com/s/iey8ka09kdm6o2g/wsjtx-2.0.0-test11.exe?dl=1
It will create a file c:\users\[username]\Appdata\Local\Temp\WSJT-X_trace.log
Note the times when decoding stops and also when you click on something that
re-enables it.
I need the debug file with those notes.
Mike
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
