Hi Mike,
Neither of the cases you describe is the result of a slow or
overburdened CPU, which is what K0VM thought he was observing.
Obviously you must have a properly functioning audio system with no
buffer underruns, etc., or decoding will be compromised.
Power-save options that cause parts of your system to "sleep" can
certainly cause problems. Properly tuning and operating a remote system
is non-trivial.
-- 73, Joe, K1JT
On 3/4/2019 15:32, Mike Lewis wrote:
Decodes do fail if the reads for the audio buffers are delayed too long and
unfortunately there is no easy way to spot this. A debug version with audio
buffer timings displayed was used to diagnose my oddball case ultimately
tracked down to the Windows advanced power option for the Display Timeout, had
to set it to Never.
Another case is if audio is briefly interrupted, the decoding and waterfall
stop, Monitor button remains green and the clock remains operational. A change
to the audio in settings or restart of WSJT-X is required to get things rolling
again. This is highly repeatable if you RDP in, start decoding, drop the
session (even quickly as an internet interruption for 1 second) and come back
to see the results. Decoding will have stopped.
In the cases above decoding either stopped or cycles missed. It would be very
helpful to put up a warning flag for these events. A display or log file
message and/or Monitor button turned red or orange. A further step would be to
attempt some self-healing such as buffer reinit after failure is detected. A
use case: I run WSJT-X remote much of the time, mostly monitoring in WSPR or
FT8 with spotting at a remote site. I have to keep the RDP session active all
the time, and cannot tolerate any internet hiccups, so cannot monitor through
the nights reliably.
73,
Mike
K7MDL
-----Original Message-----
From: Joe Taylor <j...@princeton.edu>
Sent: Monday, March 4, 2019 15:00
To: WSJT software development <wsjt-devel@lists.sourceforge.net>
Subject: Re: [wsjt-devel] Busy CPU
Hi Al,
Decodes do not fail because of a slow or overburdened CPU. They simply take
longer.
This is true for all modes supported in WSJT, MAP65, and WSJT-X.
-- 73, Joe, K1JT
On 3/4/2019 2:37 PM, Al K0VM wrote:
(WSJT-X 2.0.1)
I have noticed that at times while running FT8 or WSPR, that
decodes failed or where missed. Further observation reveals that this
can occur when the CPU it at or near 100% load. In WSPR the decode
button never highlighted. In FT8, the decode button highlighted but no
decode where displayed. ( the time was correct and signals where
displayed int the wide graph.
It's easy to say to a user 'reduce the CPU load' or 'get a bigger
CPU' but there is no clue for a new user that the decodes failed for
CPU loading.
I think it would be useful if WSJT-X could pop up a message stating
'Decode failed due to lack of CPU resources'
AL, K0VM
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://nam01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Fwsjt-devel&data=02%7C01%7C%7C8dd1b8d042364306902d08d6a0dc492b%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636873265307109935&sdata=tslgcUZaShr%2F%2B4IiZQzyykUzo%2FeYNGE%2BNUZDwq4hjwg%3D&reserved=0
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
