Hi Rich,

you describe symptoms that could be explained by your time synchronization application making step changes, that is sure to cause discontinuities in the audio streams which in turn are very likely to disrupt decoding of all signals in periods where that happens. One cause of this could be multiple time sync applications running simultaneously, e.g. not disabling the Windows Internet Time Sync tool if you are running another 3rd party time sync tool. Note also that we no longer recommend SNTP time sync tools as they are prone to this sort of time stepping on systems that keep poor time when unsynchronized. On MS Windows the Meinberg NTP Client is a proper NTP implementation and only steps the time during initialization, if configured to do so, to get a fast synchronization. There are other time sync tools that implement NTP rather than SNTP, if Meinberg is not to your liking. On other platforms the default network time sync tool will be a full NTP implementation like the *nix ntpd or more modern chronyd, so there just enabling the system network time synchronization is sufficient.

73
Bill
G4WJS.

On 19/04/2021 17:21, Rich - K1HTV wrote:
Bill - W2PKY,
Thanks for the idea of playing back the WAV files which had few or no decodes.  I played the WAV files back and they sound normal, full of signals but still decode few if any of them when I play them back for decoding.

I wonder if WSJT-X has a shared memory buffer which another program had access to which caused part of the decoding part of the program to produce the poor decodes. Another thing that I noticed was that when there were FT8 decodes on the odd cycles, at times most of the decoded signals had a DT of 0.1 then in some of the next receive cycles almost all decoded stations had a DT of 0.4 seconds. Then they would be back to 0.1 second DT numbers. However, when the problem occured, as I checked the PC clock a number of times using the "time.is <http://time.is>" URL, it indicated "Your time is Exact" usually within 1- milliseconds of exact.

My station is band switching and decoding 24/7. I've left all normally used programs running but have shut down Chrome. So far, in the past few days the problem has not re-occurred. Maybe Chrome is the culprit. The last time when the problem occured, Chrome was consuming over 1 GB of RAM, still far less than the available 16GB but still a lot. Once the problem starts, shutting down Chrome and apps doesn't fix it, only a reboot of the computer.

Thanks also for the suggestion about checking if standby memory is being used when the problem occurs. Will have to do that.

73,
Rich - K1HTV.

---------- Forwarded message ----------
From: Bill B <[email protected] <mailto:[email protected]>>
To: WSJT software development <[email protected] <mailto:[email protected]>>
Cc:
Bcc:
Date: Fri, 16 Apr 2021 21:56:27 -0400
Subject: Re: [wsjt-devel]
Hello Rich-
Have you listened to the .wav files to see if they sound clean and undistorted?

Windows allows audio channels to be monitored. Open sound control panel > recording tab, locate input device > properties > listen tab > click listen to this device. Listen to the audio quality when the app fails to decode everything that’s in the waterfall.

Bring up Task Manager > performance > open resource monitor > Memory and check if Standby memory is used up. That’s when my system starts acting strange. You may need to reboot more often.

Hope this helps.
Bill W2PKY

Sent from my iPad

On Apr 16, 2021, at 1:47 PM, Mike Lewis <[email protected] <mailto:[email protected]>> wrote:



I have been seeing cycles with only a few decode counts on busy bands periodically

I am usually run 2 WJST-X instances 24/7 comparing the decode count each cycle of a new SDR rig while I am working on it, compared to my K3, both on the same antenna.  I watch the decode count number and waterfall images constantly, but not the actual decodes much.  I do not reboot very often, the short decode count only happen to me once then back to normal. I blamed that on the possibility the missing decodes would appear in the next decode group as late decodes, but I have not verified that happens/happened. Need to look  at that next time.   I usually see it a few times a day. I will look to see if it is odd or even as well.  This is 2.4.0-rc4.  It occurred on earlier rc versions, maybe on 2.3 also.

Mike

K7MDL  EL87sm & CN88sf

*From:* Rich - K1HTV <[email protected] <mailto:[email protected]>>
*Sent:* Friday, April 16, 2021 08:14
*To:* WSJT <[email protected] <mailto:[email protected]>>
*Subject:* [wsjt-devel] Few FT8 decodes on Even cycles

In the past couple of months, using both the GA and the RC versions of WSJT-X, I've been plagued with an intermittent decode problem while running FT8. I am using Win10 on a dual core i5-2500 @ 3.3GHz with 16GB of RAM. It may take days and sometimes less for the problem to occur. On a busy 20 Meter band, the decode count and corresponding number of decodes are normally between 30 and 60 per 15 second FT8 sequence.

When the problem develops, with a full waterfall display, on the *_odd cycles_*, only 4 to 8 signals are decoded and counted. On the _even cycles_ only a few and often NONE of the many dozen signals are decoded. The DT on most of the few decoded signals is usually 0.1 seconds, so it's not a clock issue.

The only fix is to reboot the computer and then the count goes back to the 30 to 60 decodes on both even and odd 15 second cycles.

When the problem occurs, checking with Task Manager, the max CPU usage on WSJT-X peaks at less than 5% and the max total CPU usage with other apps is no more than 20% on peaks. A max of 30% of the 16GB RAM is being used. I use the Chrome browser, using 300-800MB of RAM. When I shut down Chrome and all other apps, there doesn't appear to be any phantom occurrences of their processes running, but the very poor decoding problem still exists.

I turned on WSJT-X "Save all" and played back the saved WAV files and the number of decodes matched the low counter readings noted for the same sequences. I switched from using the Elecraft K3 USB CODEC to using the sound card with audio from the K3 Line Out jack and get the same poor performance when the problem arises.

As I said, only a reboot of the computer will bring things back to normal. I'm at a loss to figure out what is causing this intermittent problem and how to correct it. Does anyone have any ideas of the cause and how it can be fixed? Thanks.

73,

Rich - K1HTV


_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel

Reply via email to