locked Re: Seeing a Lot of Skipping Decodes Every Other Sequence FT8


Hasan Schiers N0AN
 

Hi Al,
It works perfectly with SpotCollector and I  prune like you do, so the file is very small. Good suggestion, though. No failures to decode now in hours of operation when X feeds SpotCollector directly. 

It's a mystery to me.

73, N0AN
Hasan


On Mon, Mar 16, 2020, 7:01 PM Al Groff via Groups.Io <al.k0vm=yahoo.com@groups.io> wrote:
Hasan,
  You might look at the size of the spots database ...\SpotCollector\Spots.mdb. If its large then you may wish to aggressively prune it size ( throw away non-recent spots ) .. Its under spot collector> config>Spot Database tab..    Each decode that JTAlert passes to dxlab results in a spot collector lookup ( i think ). I have mine set to Prune entries older than 0.1 days and prune hourly..
FMMV
AL, K0VM

On 3/16/2020 5:26 PM, Hasan Schiers N0AN wrote:
On a busy FT8 Band like 20m, I'm seeing more and more tendencies to skip an entire sequence following a decoded sequence.

I'm on 20m FT8 now and it is happening nearly every other sequence, or it will decode 2 seq in a row and then start skipping again, without any intervention on my part.

Then it will start decoding again, do several sequences in a row (with me not transmitting, just monitoring), and then skip another sequence. I just watched it do it again while typing.

I have stopped interacting with the program, I am just letting it run with no intervention, receiving:

Two more in a row of > 20 stations
Then skipped an entire sequence
Full seq of decodes > 20
Skipped the next sequence
Full seq of decodes > 20 stations
Skipped the next seq
Full seq of decodes > 20 stations
Skipped the next seq: no decodes
Full seq of decodes > 20 stations
Skipped the next seq: no decodes
Full seq of decodes > 20 stations
Skipped the next seq.

The computer running the X software is not being touched. I'm typing on a different machine.

WSJT-X 2.12, Win10 Pro, Core I5 3.2 GHz,  8 gB of RAM (Win10 Pro is fully up to date)

While receiving CPU is about 6 %
While decoding at end of seq: 30-50%
When it skips a cycle, CPU does not go nearly as high, perhaps 25% max

UDP > JTAlert
JTAlert Rebroadcast UDP > DXLab SpotCollector


I have WSTX folder excluded from virus check
I have the WSJTX appdata folder excluded.

I just stopped JTAlert and am now having WSJT-X talk UDP directly to SpotCollector.

Every sequence is decoding > 20 stations, No Skips
CPU Utilization is 30 to 50 % as before at the decode
++++++++++++++++++++++++++++++++++++++++++
It's been about 20 minutes with absolutely perfect decoding of > 20 stations every sequence. I only made one change.

Last decode seq, the CPU went as high as 66% ...still got perfect decodes.

I have no explanation or theories to offer, but if I want consistent decodes I can't run JTAlert ...and I love JTAlert, but I'm tired of fighting this.

Hasan


    


Join main@WSJTX.groups.io to automatically receive all group messages.