The new machine at W3ENR is up and running.  Falling back to Ubuntu 20 did the trick and wsprdaemon installed smoothly.

It seems like there are a diversity of solutions to the CPU demands of FST4W. I'm not promoting mine, but will report its performance to add to the general knowledge base...

Setup:  I am decoding all modes on 630 and 160, and all modes except FST4W-1800, on all other bands up to 10 meters.

Machine: Used Dell small PC, US $275 shipped from ebay, with a I5-9500 processor.  I had a spare M2 drive to replace the old HDD it came with.

[ TLDR version:  It seems to work alright, but I don't have a lot of overhead.]

Having watched it work for a couple hours:

 - 'resting' CPU use ~18%, roughly 8% system and 10% user.

 - 'idle' CPU temps average ~39c

 - minor decoding runs (i.e. when longer FST4W isn't up) generate user demand of 30% to 40% overall CPU and core temps in the high 60s to low 70s (non climate controlled space).

 - major decoding runs generate user demand of ~80% CPU, with some cores maxxed out, and core temps in the mid-70s.

 - maximum core temp observed - only once, on one core - is 80c (82c defined as "high", 100c as "critical")

In terms of time to decode, I'm not sure what 'running out of time' exactly means, but what I've observed, judging by the CPU history graph, is:

- Most runs finish in well under 13 seconds, occasionally one or two cores will linger another second or two.

- The bigger runs at 00 and 30 minutes take about 18-22 seconds before all cores settle back down.

--

I can't place much faith in the (low) numbers that powertop is giving me.  I don't have an AC wattmeter on hand, I'll get one and report back.

EH

Thanks for the report

What is the resting and peak power consumption?

Following up on this question:

About 16w 'resting'

A brief spike of 50-80w for 'normal' decoding runs, quickly tapering off.

A spike of 110-120w for major runs (00 and 30 minutes, FST4W-1800 on 3 bands), taking a few more seconds to taper off.

EH

On Thu, Jul 7, 2022 at 7:07 AM Edward Hammond <manager@...> wrote:

The new machine at W3ENR is up and running.  Falling back to Ubuntu 20
did the trick and wsprdaemon installed smoothly.

It seems like there are a diversity of solutions to the CPU demands of
FST4W. I'm not promoting mine, but will report its performance to add to
the general knowledge base...

Setup:  I am decoding all modes on 630 and 160, and all modes except
FST4W-1800, on all other bands up to 10 meters.

Machine: Used Dell small PC, US $275 shipped from ebay, with a I5-9500
processor.  I had a spare M2 drive to replace the old HDD it came with.

[ TLDR version:  It seems to work alright, but I don't have a lot of
overhead.]

Having watched it work for a couple hours:

  - 'resting' CPU use ~18%, roughly 8% system and 10% user.

  - 'idle' CPU temps average ~39c

  - minor decoding runs (i.e. when longer FST4W isn't up) generate user
demand of 30% to 40% overall CPU and core temps in the high 60s to low
70s (non climate controlled space).

  - major decoding runs generate user demand of ~80% CPU, with some
cores maxxed out, and core temps in the mid-70s.

  - maximum core temp observed - only once, on one core - is 80c (82c
defined as "high", 100c as "critical")

In terms of time to decode, I'm not sure what 'running out of time'
exactly means, but what I've observed, judging by the CPU history graph, is:

- Most runs finish in well under 13 seconds, occasionally one or two
cores will linger another second or two.

- The bigger runs at 00 and 30 minutes take about 18-22 seconds before
all cores settle back down.

--

Rob Robinett

AI6VN

rob@...

mobile: +1 650 218 8896