I've had to do a lot of teleconferencing (Discord, WebEx, Zoom) over
the last two years. Unfortunately, none of my computers can keep up.
After a while, the CPU will go to around 300% (as shown by top),
video of other participants will freeze, and my voice will get
garbled.

Pulling the plug on my camera buys me some time, but eventually, even
with my PC not sending any video out, problems arise.

The duration varies with the platform. Worst case is Discord. Until
roughly late November, I could go for forty-five minutes or more before
I needed to disconnect my camera. Since then, noticeable voice lag
happens in under two minutes.

It's not a memory issue, at least not according to top.

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

--
Michael F. Stemper
Outside of a dog, a book is man's best friend.
Inside of a dog, it's too dark to read.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Am Freitag, 21. Januar 2022, um 13:19:11 Uhr schrieb Michael F. Stemper:

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

What hardware do you have at this time?

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 21.01.2022 20:19, Michael F. Stemper wrote:

I've had to do a lot of teleconferencing (Discord, WebEx, Zoom) over
the last two years. Unfortunately, none of my computers can keep up.
After a while, the CPU will go to around 300% (as shown by top),
video of other participants will freeze, and my voice will get
garbled.

Pulling the plug on my camera buys me some time, but eventually, even
with my PC not sending any video out, problems arise.

The duration varies with the platform. Worst case is Discord. Until
roughly late November, I could go for forty-five minutes or more before
I needed to disconnect my camera. Since then, noticeable voice lag
happens in under two minutes.

It's not a memory issue, at least not according to top.

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

I use a laptop that was standard 4+ years ago, where standard means
just an average system in all respects; CPU (i5) speed (2.5 GHz) and
cores (4), memory (8 GB), a standard built-in graphics card, etc.
Though in meetings I mostly disable video to save bandwidth. Something
that might matter is the Internet connection; recently I upgraded from something like 12/1 Mbit/s to 300/100 Mbit/s, and everything seems to
go more smoothly where other communication partners have issues. But
it's also observable quite often that the servers are overloaded and
have issues that significantly affect the quality of communication
with the clients systems, from delays, audio issues, to disconnects.
I use primarily Webex, occasionally Teams, or Zoom, and tried Jitsi.
(Note: this all is just from personal observation no hard facts.)

Janis

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 21/01/2022 13.40, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 13:19:11 Uhr schrieb Michael F. Stemper:

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

What hardware do you have at this time?

One of them is a Dell Inspiron 15n, with:
Pentium Dual Cord T4200 2.0 GHz/800 MHz/FSB/ 1 MB cache
3 GB Shared Dual Channel DDR2 at 800 MHZ
Intel Graphics Media Accelerator X4500HD
Dell Wireless 1397 802.11g Half Mini-card
WD 500 GB SATA 6 Gb/s, 7200 rpm, 32 MB cache (not original)
(Information from invoice)

The other is a Dell Latitude E6410, with:
2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
Cache:
L1: 32 KiB
L2: 512 KiB
L3: 3 MiB
Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock
Broadcom BCM43224 802.11a/b/g/n
SCSI disk: 57 GiB
(Info from lshw, as I wasn't able to snapshot the invoice when
I ordered this one.)

These don't work with teleconferencing on Ubuntu. What should I
buy that will?


--
Michael F. Stemper
Psalm 82:3-4

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Michael F. Stemper wrote:

On 21/01/2022 13.40, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 13:19:11 Uhr schrieb Michael F. Stemper:

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

What hardware do you have at this time?

One of them is a Dell Inspiron 15n, with:
  Pentium Dual Cord T4200 2.0 GHz/800 MHz/FSB/ 1 MB cache
  3 GB Shared Dual Channel DDR2 at 800 MHZ
  Intel Graphics Media Accelerator X4500HD
  Dell Wireless 1397 802.11g Half Mini-card
  WD 500 GB SATA 6 Gb/s, 7200 rpm, 32 MB cache (not original)
(Information from invoice)

The other is a Dell Latitude E6410, with:
  2 x Intel i5 CPU,Dell Latitude E6410, 2.53 GHz, 64 bits, 533 MHz clock
  Cache:
   L1:  32 KiB
   L2: 512 KiB
   L3:   3 MiB
  Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock
  Broadcom BCM43224 802.11a/b/g/n
  SCSI disk: 57 GiB
(Info from lshw, as I wasn't able to snapshot the invoice when
I ordered this one.)

These don't work with teleconferencing on Ubuntu. What should I
buy that will?

I have no issues with my old Thinkpad L530 with about the same vintage
CPU i5-3230M CPU @ 2.60GHz 8GB RAM 3rd Gen Core processor Graphics
Controller. Had done some Zoom meetings, but prefer and use mostly
Google Meet.

--
Take care,

Jonathan
-------------------
LITTLE WORKS STUDIO
http://www.LittleWorksStudio.com

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/21/22 12:22, Janis Papanagnou wrote:

On 21.01.2022 20:19, Michael F. Stemper wrote:

I've had to do a lot of teleconferencing (Discord, WebEx, Zoom) over
the last two years. Unfortunately, none of my computers can keep up.
After a while, the CPU will go to around 300% (as shown by top),
video of other participants will freeze, and my voice will get
garbled.

Pulling the plug on my camera buys me some time, but eventually, even
with my PC not sending any video out, problems arise.

The duration varies with the platform. Worst case is Discord. Until
roughly late November, I could go for forty-five minutes or more before
I needed to disconnect my camera. Since then, noticeable voice lag
happens in under two minutes.

It's not a memory issue, at least not according to top.

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

I use a laptop that was standard 4+ years ago, where standard means
just an average system in all respects; CPU (i5) speed (2.5 GHz) and
cores (4), memory (8 GB), a standard built-in graphics card, etc.
Though in meetings I mostly disable video to save bandwidth. Something
that might matter is the Internet connection; recently I upgraded from something like 12/1 Mbit/s to 300/100 Mbit/s, and everything seems to
go more smoothly where other communication partners have issues. But
it's also observable quite often that the servers are overloaded and
have issues that significantly affect the quality of communication
with the clients systems, from delays, audio issues, to disconnects.
I use primarily Webex, occasionally Teams, or Zoom, and tried Jitsi.
(Note: this all is just from personal observation no hard facts.)

Janis

I don't use Zoom but depend on Jit.se meeting for SF-LUG virtual meetings. I have used my old E6540 but mainly use my newer E7450.
Both are i7 and the E66540 has 4 cores and can handle 8 threads.
The E7450 has only 2 cores but can handle 4 threads. I would go
for a newer i7 4 core at a minimum or a advanced Ryzen 5000 and
above, Minimum of 16 GB of Ram, SSD drive and later Intel intergrated
graphics but AMD graphics integrated in Rtzen should handle it.

Just my humble opinion.

bliss - brought to you by the power and ease of PCLinuxOS
and a minor case of hypergraphia
--
bliss dash SF 4 ever at dslextreme dot com

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/21/2022 3:56 PM, Michael F. Stemper wrote:

On 21/01/2022 13.40, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 13:19:11 Uhr schrieb Michael F. Stemper:

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

Thanks for any help that you can provide.

What hardware do you have at this time?

One of them is a Dell Inspiron 15n, with:
  Pentium Dual Cord T4200 2.0 GHz/800 MHz/FSB/ 1 MB cache
  3 GB Shared Dual Channel DDR2 at 800 MHZ
  Intel Graphics Media Accelerator X4500HD
  Dell Wireless 1397 802.11g Half Mini-card
  WD 500 GB SATA 6 Gb/s, 7200 rpm, 32 MB cache (not original)
(Information from invoice)

The other is a Dell Latitude E6410, with:
  2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
  Cache:
   L1:  32 KiB
   L2: 512 KiB
   L3:   3 MiB
  Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock
  Broadcom BCM43224 802.11a/b/g/n
  SCSI disk: 57 GiB
(Info from lshw, as I wasn't able to snapshot the invoice when
I ordered this one.)

These don't work with teleconferencing on Ubuntu. What should I
buy that will?

https://ark.intel.com/content/www/us/en/ark/products/43544/intel-core-i5540m-processor-3m-cache-2-53-ghz.html

"Processor Graphics

Processor Graphics Intel HD Graphics for Previous Generation Intel Processors
"

What does that even mean ?

*******

Some GPUs (Graphic Processing Units), they have video decoders.

At one time, acceleration consisted of IDCT (inverse discrete cosine transform),
which is part of a calculation for macroblocks (square groups of pixels, to be analyzed in the frequency domain and some information thrown away to give
data rate compression).

That used to amount to little practical acceleration, as it was only
a small part of the computational load. Developers are such snobs today,
they will no longer even *touch* IDCT on video cards. Apparently, a
modern processor can compute what that function did, faster in the core,
than to prep an accelerator block and do it that way.

Later, NVidia Purevideo and AMD AVIVO/UVD/VCE handled a lot more
of the video decode functions. Now, the GPU can do practically the
whole decoding job. Maybe MPEG2 codecs or MP4 containers are used, and
the GPU can decode the video stream without a lot of help. I tested
the other day, and my video card, for example, can decode around
1100 frames a second. Intel QuickSync can decode at least five
HD streams at the same time (a claim for the first generation).

Having such functions, reduces the CPU work needed to put the video
conference on your screen.

Firefox might be using the GPU to render individual font characters.

Firefox "composites" screen renders 60 times a second, The OS
also uses compositing (all OSes have this now), so that when
a window is exposed, it does not have to be redrawn, because
the info was always present, but hidden by Z-axis priority.
The GPU does the heavy lifting (Z-axis priority) for this.

When your GPU is "too old" for this stuff, that is when fallback
code on the CPU does all the work instead.

If your CPUs had modern GPU (video card) chips, then you might
not be railed while conferencing. But never discount the
possibility of bad coding, such that even high end laptop owners
are seeing the same thing. A memory leak in CPU code, can sometimes
throw a program into the weeds (it keeps calling the OS and trying
to malloc stuff).

*******

I call this practice, when buying computers, the "buzzword compliance"
aspect of buying. For example, there are lots of new acronyms on recent
GPUs, that don't do anything I particularly care about, yet I know when
a developer seems them, some of them will become "best buddies". And
when a feature, like some AI trick, becomes popular, then all the
old GPUs that don't have the useless feature, they'll suddenly be
that much further behind.

This is how GPUs have moved from "mere frame buffers", to
"drivers of computer sales". The more GPUs you can leave
in the dust, the more new laptops you can sell.

This also means, for a prospective purchaser, they can't
get too excited about the "CPU clock rate" or the "core count",
as the GPU is sitting there thumbing its nose at you. You have
to buy GPUs that are brand new, and have the maximum number
of tick boxes ticked. And if you see review articles, if
the article says "this GPU is missing half the decoders, to
save silicon area", you know right away what they're saying
is "stay away from this piece of crap". There is one brand new
product, announced in the last two days, where the disparaging
comments are coming in about it. And the shocker is, the chip
had apparently been designed for laptops in the year 2020.
Well, no laptop owner wants half-a-job silicon either, thanks...
It's not normal for that to happen today. Shouldn't be happening.

*******

You can't shop by saying "well, Dell made it, it must be OK", because
every once in a while, the silicon companies make a "mistake". It's
your job as a consumer, to read the reviews and catch them at their
deceit.

Paul

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Am Freitag, 21. Januar 2022, um 14:56:42 Uhr schrieb Michael F. Stemper:

One of them is a Dell Inspiron 15n, with:
Pentium Dual Cord T4200 2.0 GHz/800 MHz/FSB/ 1 MB cache
3 GB Shared Dual Channel DDR2 at 800 MHZ
Intel Graphics Media Accelerator X4500HD

For me that seems to be too slow.

The other is a Dell Latitude E6410, with:
2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
Cache:
L1: 32 KiB
L2: 512 KiB
L3: 3 MiB
Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock

I wonder why that is also too slow. What about the RAM usage during the conference?

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 22/01/2022 04.45, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 14:56:42 Uhr schrieb Michael F. Stemper:

One of them is a Dell Inspiron 15n, with:
Pentium Dual Cord T4200 2.0 GHz/800 MHz/FSB/ 1 MB cache
3 GB Shared Dual Channel DDR2 at 800 MHZ
Intel Graphics Media Accelerator X4500HD

For me that seems to be too slow.

So should I be looking for something running at 3-4 GHz?

The other is a Dell Latitude E6410, with:
2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
Cache:
L1: 32 KiB
L2: 512 KiB
L3: 3 MiB
Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock

I wonder why that is also too slow. What about the RAM usage during the conference?

From memory (obviously, I can't test without a meeting), top is
generally showing 15% to 20% memory usage when CPU is 250% to 300%.

This strikes me as strange. The behavior of performance gradually
deteriorating smacks of a memory leak, but the numbers tell another
story. (Plus, it's the same behavior across Discord, WebEx, and Zoom. Presumably, they don't all have undetectable memory leaks.)


--
Michael F. Stemper
The FAQ for rec.arts.sf.written is at <http://leepers.us/evelyn/faqs/sf-written.htm>
Please read it before posting.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/22/2022 9:20 AM, Michael F. Stemper wrote:

On 22/01/2022 04.45, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 14:56:42 Uhr schrieb Michael F. Stemper:

One of them is a Dell Inspiron 15n, with:
    Pentium Dual Cord T4200 2.0 GHz/800 MHz/FSB/ 1 MB cache
    3 GB Shared Dual Channel DDR2 at 800 MHZ
    Intel Graphics Media Accelerator X4500HD

For me that seems to be too slow.

So should I be looking for something running at 3-4 GHz?

The other is a Dell Latitude E6410, with:
    2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
    Cache:
     L1:  32 KiB
     L2: 512 KiB
     L3:   3 MiB
    Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock

I wonder why that is also too slow. What about the RAM usage during the
conference?

From memory (obviously, I can't test without a meeting), top is
generally showing 15% to 20% memory usage when CPU is 250% to 300%.

This strikes me as strange. The behavior of performance gradually deteriorating smacks of a memory leak, but the numbers tell another
story. (Plus, it's the same behavior across Discord, WebEx, and Zoom. Presumably, they don't all have undetectable memory leaks.)

Your new laptop, should have a good, well researched, GPU on it.
Not all laptop GPUs have good feature sets.

The CPU speed after solving the GPU problem, is a secondary problem.
If the machine is decently new, it will have DDR4 memory, like
DDR4-3200 or so, so that won't be a problem.

AMD and Intel have been so engrossed on beating one another
over the head with 224W 5GHz+ processors for so long, it's
hard to find a good modern laptop processor. I checked a couple
tablets a few minutes ago, and one of them, from the year 2020,
was using a year 2015 processor??? What the hell?

There's even been a few desktop boxes, that instead of using
Alder Lake, they're using older processors.

Paul

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

"Michael F. Stemper" <michael.stemper@gmail.com> writes:

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

On my 2013-vintage Lenovo E130 (Core i3-3227U 1900MHz, dual-core,
integrated Intel graphics) with Ubuntu 18.04 Zoom worked nicely, as
well as BigBlueButton and (with the usual disconnect issues) Jitsi.
Never used Discord or WebEx for teleconferencing on that machine.

On my new shiny laptop (Fujitsu U7311) with Ubuntu 21.04 Zoom crashes
as soon as, e.g., I look at the participants list, but BigBlueButton
still works nicely.

Anyway, given the Lenovo E130 experience, it seems to me that
teleconferencing does not have high hardware requirements, and that
your problems may stem from software problems (like apparently my Zoom problems). I have no good suggestion on how to fix that, though.

- anton
--
M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 22/01/2022 14.37, Paul wrote:

On 1/22/2022 9:20 AM, Michael F. Stemper wrote:

On 22/01/2022 04.45, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 14:56:42 Uhr schrieb Michael F. Stemper:

For me that seems to be too slow.

So should I be looking for something running at 3-4 GHz?

The other is a Dell Latitude E6410, with:
    2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
    Cache:
     L1:  32 KiB
     L2: 512 KiB
     L3:   3 MiB
    Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock

I wonder why that is also too slow. What about the RAM usage during the
conference?

This strikes me as strange. The behavior of performance gradually
deteriorating smacks of a memory leak, but the numbers tell another
story. (Plus, it's the same behavior across Discord, WebEx, and Zoom.
Presumably, they don't all have undetectable memory leaks.)

Your new laptop, should have a good, well researched, GPU on it.
Not all laptop GPUs have good feature sets.

How do I tell whether the feature sets of any particular GPU are good
or bad? I don't know squat about hardware, so I have no idea what to
look for.

Thanks.

--
Michael F. Stemper
Always remember that you are unique. Just like everyone else.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 22/01/2022 18.18, Anton Ertl wrote:

"Michael F. Stemper" <michael.stemper@gmail.com> writes:

What kind of laptop do I need to buy in order to teleconference
with Ubuntu? Do I need a certain minimum clock speed? A GPU?
A certain number of cores (whatever they are)?

On my 2013-vintage Lenovo E130 (Core i3-3227U 1900MHz, dual-core,
integrated Intel graphics) with Ubuntu 18.04 Zoom worked nicely, as
well as BigBlueButton and (with the usual disconnect issues) Jitsi.
Never used Discord or WebEx for teleconferencing on that machine.

Well, that's older than one of my problematic boxes, so this is
good input.

On my new shiny laptop (Fujitsu U7311) with Ubuntu 21.04 Zoom crashes
as soon as, e.g., I look at the participants list,

Oy!

Anyway, given the Lenovo E130 experience, it seems to me that teleconferencing does not have high hardware requirements, and that
your problems may stem from software problems (like apparently my Zoom problems).

I'm skeptical of this idea, since it would need to be a softare
issue that affects three different platforms.


--
Michael F. Stemper
Always remember that you are unique. Just like everyone else.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

On 22/01/2022 14.37, Paul wrote:

On 1/22/2022 9:20 AM, Michael F. Stemper wrote:

On 22/01/2022 04.45, Marco Moock wrote:

Am Freitag, 21. Januar 2022, um 14:56:42 Uhr schrieb Michael F. Stemper:

For me that seems to be too slow.

So should I be looking for something running at 3-4 GHz?

The other is a Dell Latitude E6410, with:
    2 x Intel i5 CPU, M 540, 2.53 GHz, 64 bits, 533 MHz clock
    Cache:
     L1:  32 KiB
     L2: 512 KiB
     L3:   3 MiB
    Memory: 2 x 2 GiB, 64 bits, 1067 MHz clock

I wonder why that is also too slow. What about the RAM usage during the >>>> conference?

This strikes me as strange. The behavior of performance gradually
deteriorating smacks of a memory leak, but the numbers tell another
story. (Plus, it's the same behavior across Discord, WebEx, and Zoom.
Presumably, they don't all have undetectable memory leaks.)

Your new laptop, should have a good, well researched, GPU on it.
Not all laptop GPUs have good feature sets.

How do I tell whether the feature sets of any particular GPU are good
or bad? I don't know squat about hardware, so I have no idea what to
look for.

Thanks.

You run into occasional articles like this. This is a recent (few days ago) release of a desktop video card, where the chip used was actually intended
as a laptop accelerator (separate from CPU). The chip should not have
been sold in this format. Or cranked, to this electricity level :-)

https://hothardware.com/news/amd-confirms-navi-24-was-made-for-ryzen-6000-laptops

"The just-released Radeon RX 6500 XT is a weird product in a lot of ways.
Besides its lean 64-bit memory bus, its hilariously-narrow PCIe x4 interface,
and the resulting mediocre performance, it also suffers from

limited video encoding and decoding capabilities

compared to its fellow RDNA 2 GPU siblings. On top of all that, it's not
even particularly efficient, drawing some 100 watts while gaming.

The weaknesses of the card have made many wonder exactly why it is the
way it is. As it happens, there's a surprisingly reasonable answer.
Posting on the Phoronix forums, an AMD developer named John Bridgman,
formerly known as "the AMD Linux guy," acknowledged that the Navi 24 GPU
which powers the RX 6500 XT was never really meant to be a desktop graphics card.

Commenting on a post by "Hibbelharry" complaining about the
weakened media capabilities in Navi 24, Bridgman said "The primary use of
Navi24 will be in laptops paired with a Rembrandt APU, which has full video
functionality and Gen4 PCIe." A laptop so-equipped can use Rembrandt's
built-in RDNA2 graphics for video encoding and decoding (including the AV1 codec),
while the Navi 24 GPU comes into play for games and other 3D graphics.
"

NVidia offers tables like this.

https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new

Look at the table at the end of this article. It shows you that you'll
be wanting "VA-API" support, so Firefox and Chrome can accelerate Youtube playback.

https://wiki.archlinux.org/title/Hardware_video_acceleration

But a lot of the measurement side of things ("checking your PC for stuff")
is decidedly mushy and a waste of time.

Take my experience with the AMD 5600G the other day here. I installed
LinuxMint 20.3 Uma on the machine, and video is not accelerated. The
Xorg AMDGPU package seems to be installed. But it was obvious by the
way that windows would stutter when I was moving them, that compositing
was not working, and basically there was no GPU acceleration whatsoever.
Even though the CPU was faster than my previous one, the CPU simply
could not hide the deficiencies present.

I had to bump the kernel version, and magically, everything started
to work :-)

That's an example of what brand-new hardware gets you. You might
grab what would even be considered a "recent" distro, and still
have to do your homework to get it tuned up (HWE kernel).

*******

This is a dump of some recent positioning info, plus the
info for my 5600G processor. Intel would have its own
tables (if you can find them).

AMD 5600G = Cezanne (mobile)

https://en.wikipedia.org/wiki/Video_Coding_Engine ==> Cezanne is VCN 2.2

https://en.wikipedia.org/wiki/Video_Core_Next ==> Copy the VCN 2.2 row

H.262 (MPEG-2) Decode Yes
MPEG-4 Decode Yes
VC-1/ WMV 9 Decode Yes
H.264 (MPEG-4 AVC) Decode/Encode Yes
H.265 (HEVC) Decode/Encode Yes
VP9 Decode Yes
AV1 Decode No <=== silly me, I bought obsolete hardware...
JPEG Decode Yes (MJPEG??? or actual JPEG ???)

https://www.phoronix.com/scan.php?page=news_item&px=AV1-Decode-For-AMD-VCN-3.0

15 September 2020

"It turns out the Radeon RX 6000 series will have AV1 hardware video decode capabilities.

In addition to Intel Xe / Tigerlake and the NVIDIA GeForce RTX 30 series
supporting AV1 hardware decoding, it's now firmed up that the next-gen Navi 2
GPUs will also have AV1 decode."

This has some pointers to expected chip content. https://www.tomshardware.com/news/amd-van-gogh-apus-spotted-in-linux-patch-ddr5-navi-2-igpu

"The current wisdom is that the Van Gogh APUs will slot into the 7.8W to 18W
space for ultra-thin mobile devices to compete against Intel's Y-series
Tiger Lake chips. Intel's chips currently support DDR4/LPDDR4, though
Intel says Tiger Lake will support DDR5 in the future.

AMD hasn't made any official announcements detailing the Van Gogh chips,
but they're predicted to land in 2021, while AMD's Cezanne APUs with the
Zen 3 microarchitecture are thought to slot in for higher-performance applications.
"

You don't want DDR5 unless it runs at DDR5-6400 or so.
From a clock rate perspective, it's the same as DDR4-3200.
But DDR5 has two channels per DIMM/SODIMM, which potentially
could be more agile... if someone knew how to design a processor.

*******

If I try to test for support, this site seems to be coded
incorrectly for H.265. Or, maybe it's a profile issue, where
the hardware profile support does not match the profile they
are testing with.

http://html5test.com/index.html

Video codecs (5600G) Nvidia GTX1080 ff78ESR

MPEG-4 ASP support No No
H.264 support Yes Yes
H.265 support No No <=== this should actually be yes, yes
Ogg Theora support Yes Yes
WebM with VP8 support Yes Yes
WebM with VP9 support Yes Yes

I get all six boxes ticked for this one, and I used to have
trouble passing this on my now-dead desktop PC. Today, this
isn't much of a test, but it's still a good indicator for Youtube use.

https://web.archive.org/web/20190102014458/https://www.youtube.com/html5

The FFMPEG info dump, was indicative, but since an AMD processor
does not support NVENC or NVDEC, some of the entries FFMPEG dumps
are not "tested" or "loaded" or "runtime ready" indicators, and
are more static indicators of what is compiled in.

And this sort of info, is not helpful either, because the
browser may use a fallback code path, and when buying a new
computer, you don't want it to do that.

https://support.mozilla.org/en-US/kb/html5-audio-and-video-firefox

HTML5 audio and video in Firefox...

*******

This may help a bit with NVidia support for video formats. My GTX1080
will burn up 60W or electricity, but only when decoding at a rate of
1100 frames per second (thirty plus videos worth...). The actual power consumption for a single video, is unmeasureably low on the graph.
The tabular nvidiasmi can dump a number you can read a bit easier.

https://i.stack.imgur.com/5nvHv.png

This gives some idea what to expect when you buy NVidia...

https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new

But since Intel is likely to be in your laptop, we need to find a table for it. The part I copied here, could apply to Gen10/Gen11/Gen12-alderlake silicon.

https://en.wikipedia.org/wiki/Intel_Quick_Sync_Video

"Version 8 (Tiger Lake, Rocket Lake, Alder Lake, Raptor Lake)

The Tiger Lake (microprocessor), Rocket Lake, Alder Lake & Raptor Lake adds
VP9 12-bit & 12-bit 4:4:4 hardware decoding and
HEVC 12-bit 4:2:0, 4:2:2 and 4:4:4 hardware decoding.
Gen12 Xe will also support native AV1 decode,
which includes 10-bit 4:2:0 16K stills and 10-bit 4:2:0 8K,
4K and 2K video.

Hardware encoding for VP8 was dropped and hardware decoding
is only available on Tiger Lake.
"

It would be better if that was rendered as a table, but if you
look at the complexity of the AMD detailed table, I can see
why a Wikipedia author would not want to bother.

There's probably no reason today to worry about AV1 decoding,
but, on the other hand, you don't want to buy obsolete hardware
if you can avoid it.

Paul

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

On 22/01/2022 14.37, Paul wrote:

I appreciate you taking the time to write all of that up, but
most of it was right over my head. I really don't know anything
about hardware.

I think that your main point was that I should read reviews of
the graphics card before buying something.

The first box that I looked at was: <https://www.dell.com/en-us/shop/cty/pdp/spd/inspiron-15-3511-laptop#configurations_section>
It had:
- Up to 11th Generation Intel(R) Core[tm] i7-1165G7 Processor
- Up to Intel(R) Iris(R) Xe Graphics with shared graphics memory

The first non-Intel page on this one: <https://laptoping.com/gpus/product/intel-iris-xe-g7/>
says things like:
- "[...] belongs to the entry-level graphics processor category."
- "[...] since they don’t have their own video memory, the Iris
Xe can’t compete against the full-blown gaming-class video cards."
These make it sound as if buying a box with this won't help with
my issues.

On the other hand, they also say:
- "[...] performance of the Iris Xe is similar to performance of
some entry-level dedicated video cards like the Nvidia GeForce
MX series."
which makes it sound all right.

Then, I found: <https://www.dell.com/en-us/shop/gaming-laptops/g15-ryzen-edition-gaming-laptop/spd/g-series-15-5515-laptop/gn5515eytwh>
which has
- AMD Ryzen[tm] 5 5600H 6-core/12-thread Mobile Processor
- NVIDIA(R) GeForce RTX[tm] 3050 4GB GDDR6

This page: <https://www.pcworld.com/article/608294/nvidia-geforce-rtx-3050-review.html> says:
- "Great 1080p gaming with high visual settings"
But, is what's good for gaming automatically good for teleconferencing?

Is it worth twice as much as the first? Obviously, if the first one won't do teleconferencing, its value to me is nothing. But, I don't want to spend an extra $400 if it's not necessary.

--
Michael F. Stemper
A preposition is something you should never end a sentence with.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/28/2022 3:52 PM, Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

On 22/01/2022 14.37, Paul wrote:

I appreciate you taking the time to write all of that up, but
most of it was right over my head. I really don't know anything
about hardware.

I think that your main point was that I should read reviews of
the graphics card before buying something.

The first box that I looked at was: <https://www.dell.com/en-us/shop/cty/pdp/spd/inspiron-15-3511-laptop#configurations_section>
It had:
- Up to 11th Generation Intel(R) Core[tm] i7-1165G7 Processor
- Up to Intel(R) Iris(R) Xe Graphics with shared graphics memory

The first non-Intel page on this one: <https://laptoping.com/gpus/product/intel-iris-xe-g7/>
says things like:
- "[...] belongs to the entry-level graphics processor category."
- "[...] since they don’t have their own video memory, the Iris
  Xe can’t compete against the full-blown gaming-class video cards." These make it sound as if buying a box with this won't help with
my issues.

On the other hand, they also say:
- "[...] performance of the Iris Xe is similar to performance of
  some entry-level dedicated video cards like the Nvidia GeForce
  MX series."
which makes it sound all right.

Then, I found: <https://www.dell.com/en-us/shop/gaming-laptops/g15-ryzen-edition-gaming-laptop/spd/g-series-15-5515-laptop/gn5515eytwh>
which has
- AMD Ryzen[tm] 5 5600H 6-core/12-thread Mobile Processor
- NVIDIA(R) GeForce RTX[tm] 3050 4GB GDDR6

This page: <https://www.pcworld.com/article/608294/nvidia-geforce-rtx-3050-review.html> says:
- "Great 1080p gaming with high visual settings"
But, is what's good for gaming automatically good for teleconferencing?

Is it worth twice as much as the first? Obviously, if the first one won't do teleconferencing, its value to me is nothing. But, I don't want to spend an extra $400 if it's not necessary.

1165-G7

https://community.intel.com/t5/Graphics/Does-Intel-Iris-Xe-decode-encode-HEVC-H265-4-2-2-10-Bit-Video/td-p/1272635

"11th Generation Intel Core Processors, such as Intel Core i7-1185G7"

https://community.intel.com/cipcp26785/attachments/cipcp26785/graphics/96220/1/631121-004%20(1).pdf

If you have trouble reading that, try this to make a filtered version, with perhaps a little less PDF wizardry in it. This seemed to work for me. You want to
go to Page 104 in any case. I can read it now, but I can't copy and paste
(it's using the font translation hack).

mutool convert -F pdf -O decompress,clean -o filtered.pdf 631121-004.pdf

In any case, there is a table of video decoding formats there.

For encode, you can see the cupboard is a little less full.

H264, H265, VP9

There are just a few GPU designs that include VP8 and VP9
at the same time. But I suppose you could say "VP9 is the future"
and leave it at that. In any case, I think the H264 should help.

*******

AMD Ryzen[tm] 5 5600H 6-core/12-thread Mobile Processor
NVIDIA(R) GeForce RTX[tm] 3050 4GB GDDR6

I don't even know if Linux has a BumbleBee for this combination.
Normally, it would be an Intel GPU and an NVidia one, the NVidia one
would write images into the frame buffer of the Intel GPU. I suppose
this does not require a lot of sophistication in the handling of the
CPU frame buffer, to make it work.

When doing Zoom, I would expect the session to run on the 5600H and
its frame buffer. The NVidia should remain at 100MHz and idling
(low power consumption). Thus, we need to know which version of
VCE this is. And my 5600G did not have the latest VCE.

https://en.wikipedia.org/wiki/Video_Coding_Engine

Follow down from "Cezanne" at the top of the diagram, gives VCN 2.2

https://en.wikipedia.org/wiki/Video_Core_Next

The VCN 2.2 row is missing AV1

I don't really know if BumbleBee/Optimus uses video decoders
off the high-power GPU or not. If that were the case, you'd then
need the NVidia table.

And the NVidia table doesn't have the 3050 yet. And you can
see there is an attitude problem at NVidia. Customers have to
drag the information out of them.

https://forums.developer.nvidia.com/t/video-encode-and-decode-gpu-support-matrix/64780/61

My concern was, the GPU silicon part number is different than the
other RTX family members (as you'd expect), which gives an opportunity
for a different support table for video decode.

Paul

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

"Michael F. Stemper" <michael.stemper@gmail.com> writes:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

On 22/01/2022 14.37, Paul wrote:

I appreciate you taking the time to write all of that up, but
most of it was right over my head. I really don't know anything
about hardware.

I think that your main point was that I should read reviews of
the graphics card before buying something.

The first box that I looked at was: ><https://www.dell.com/en-us/shop/cty/pdp/spd/inspiron-15-3511-laptop#configurations_section>
It had:
- Up to 11th Generation Intel(R) Core[tm] i7-1165G7 Processor
- Up to Intel(R) Iris(R) Xe Graphics with shared graphics memory

The first non-Intel page on this one: ><https://laptoping.com/gpus/product/intel-iris-xe-g7/>
says things like:
- "[...] belongs to the entry-level graphics processor category."
- "[...] since they don’t have their own video memory, the Iris
Xe can’t compete against the full-blown gaming-class video cards."
These make it sound as if buying a box with this won't help with
my issues.

Lots of people have laptops with this or weaker graphics, and the
video conferencing software developers certainly targeted them rather
than requiring high-end gaming systems. As mentioned, my old laptop
was much weaker and Zoom worked just fine.

- anton
--
M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 29.01.2022 08:24, Anton Ertl wrote:

Lots of people have laptops with this or weaker graphics, and the
video conferencing software developers certainly targeted them rather
than requiring high-end gaming systems. As mentioned, my old laptop
was much weaker and Zoom worked just fine.

This matches my experience and I wondered about some statements from
the high-end advocates here.

My expectation would also be that if only low-bandwidth is available
or CPU/GPU performance is low that tools would reduce the quality of
the video (less frames or lower resolution or something like that) -
if necessary in the first place -, isn't that the case?

As mentioned in my other post, bandwidth and server-stability seemed
more an issue (add software-quality in some cases, depending on the
used tools). Also if your client is overloaded with lots of parallel
processes that consume many resources (time and memory).

Janis

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 1/29/2022 4:46 AM, Janis Papanagnou wrote:

On 29.01.2022 08:24, Anton Ertl wrote:

Lots of people have laptops with this or weaker graphics, and the
video conferencing software developers certainly targeted them rather
than requiring high-end gaming systems. As mentioned, my old laptop
was much weaker and Zoom worked just fine.

This matches my experience and I wondered about some statements from
the high-end advocates here.

My expectation would also be that if only low-bandwidth is available
or CPU/GPU performance is low that tools would reduce the quality of
the video (less frames or lower resolution or something like that) -
if necessary in the first place -, isn't that the case?

As mentioned in my other post, bandwidth and server-stability seemed
more an issue (add software-quality in some cases, depending on the
used tools). Also if your client is overloaded with lots of parallel processes that consume many resources (time and memory).

Janis

The OP could also change OSes and retest.

Paul

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

On 22/01/2022 14.37, Paul wrote:

I appreciate you taking the time to write all of that up, but
most of it was right over my head. I really don't know anything
about hardware.

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old
(ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it
couldn't do anything fancy like a virtual background or anything, but
that wasn't super important.

[...]
- "[...] belongs to the entry-level graphics processor category."
- "[...] since they don’t have their own video memory, the Iris
Xe can’t compete against the full-blown gaming-class video cards."
These make it sound as if buying a box with this won't help with
my issues.

No, you're reading far too much into it. Only thing they mean is that
it's not gonna compete with the top-end monsters that nVidia and AMD put
out. Which is perfectly fine - you don't need some $1000 monster card to
join a zoom conference.

[...]
- "Great 1080p gaming with high visual settings"
But, is what's good for gaming automatically good for teleconferencing?

It depends on what you mean by "good for". An nVidia 3050 is certainly
capable of it ...

Is it worth twice as much as the first? Obviously, if the first one
won't do teleconferencing, its value to me is nothing. But, I don't
want to spend an extra $400 if it's not necessary.

The extra $400 is because you're getting one of those nVidia monster
cards. And yes, you're just throwing money away if all you need is "conferencing".


-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE3asj+xn6fYUcweBnbWVw5UznKGAFAmH3vS8ACgkQbWVw5Uzn KGDC4hAArqTmfwiCMhi4Ca2x6j9ieahIkKvUnIX3J/orZkbEWhZdJ50ZTNrNJf3I BsR+lolEegCDhZ5IfLfRw102e5bu33yfAfGXdiS0RhY9y6/RPI+xhyDLAzlPQ39t 3H03zP82CzLOIHJVqxe6ztm9kiYyFyCCDLBLaTP6ldyVWWDHraRiLIGsSkkjOM36 NlnjRHwocAVrnY1ai6mcWuVUYjtBFN7mx9ShGSQF/mrYSqFUTui7XHdd6CPZibkK LU5bwzaKMvqs0glbKoWgwiJ8O2XtCyHciqcJaa2/BnomVOsKxFW+oOxIFDuHuZq/ Q9NUx3f7rDE2+Cj/cB4eaDaEJWyc89yxOuTIFv8g6KZOIaX+W/2xnRFeaPInaQ5M F3XEWEV44EuCxTGCGaRdvnciNSMypUtKMqE45vc7Xc4qK+7TjmMbwxUydQ/rbdDu 4PucMzySj1oUAnUWc9//Y9b4UvzYP1/4QqZ07iIR9POdVYlxRP5cMQLsOQJxpAoI 0xJvF7UDpA4gAlacPHJYN6VbC+q5eyw9wCX7U5GJvxD5/mIGV3Dh340EK5Q1ijOS eEgymSQoAWqHsK52mbK/d17wrYsCd6Vuro7qqjVPqcKi4IP5qmOED576SEZkC7dh 0zVY8faQy0QAygqAOxbzH8qi/cq0plgVVhlT/1WS9mqdoHUsahI=
=M52R
-----END PGP SIGNATURE-----

--
|_|O|_| Github: https://github.com/dpurgert
|_|_|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
|O|O|O|

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old
(ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it
couldn't do anything fancy like a virtual background or anything, but
that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based Latitude, neither of which can keep up with Discord/WebEx/Zoom.

- "[...] belongs to the entry-level graphics processor category."
- "[...] since they don’t have their own video memory, the Iris
Xe can’t compete against the full-blown gaming-class video cards."
These make it sound as if buying a box with this won't help with
my issues.

No, you're reading far too much into it. Only thing they mean is that
it's not gonna compete with the top-end monsters that nVidia and AMD put
out. Which is perfectly fine - you don't need some $1000 monster card to
join a zoom conference.

Okay. Didn't really want to spend the extra money. On the other hand, didn't want to buy a computer to solve a specific problem that ended up not solving the problem.

[...]
- "Great 1080p gaming with high visual settings"
But, is what's good for gaming automatically good for teleconferencing?

It depends on what you mean by "good for". An nVidia 3050 is certainly capable of it ...

But, if I understand you correctly, it's serious overkill for my purposes.

Is it worth twice as much as the first? Obviously, if the first one
won't do teleconferencing, its value to me is nothing. But, I don't
want to spend an extra $400 if it's not necessary.

The extra $400 is because you're getting one of those nVidia monster
cards. And yes, you're just throwing money away if all you need is "conferencing".

I guess that I'll go ahead and buy the first one. Thanks.

--
Michael F. Stemper
A preposition is something you should never end a sentence with.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old
(ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it
couldn't do anything fancy like a virtual background or anything, but
that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these
machines have? If it's 4GiB or less, it's a lot more likely that's your
problem (or rather, the heavy swapping that'd be necessary to run).


-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE3asj+xn6fYUcweBnbWVw5UznKGAFAmH6s1oACgkQbWVw5Uzn KGDryg//WLn4vldbGVndyJ7XusHzIBdta7ufwp1ujk7MVQfZLNeq0IpPaEumJWH7 6h44vAQsOHDQcEl/Zq7kwaF1KrzgMctqa+zOwLmEXcIjmLF6CoP0YfXIn0g7Kznz 7jOvbMq5EvdiCV6gXMR4PYdLK2gjtbwVJq8eRGsVFb2fH2tTX1l+vHjNE8MlXRrL fppm0/BQTs/dYtMaoavsSUOko3Xunc3GrGYP94BQlajjB16kTnTqT8umGft38que nwS723s2/mkuCpk9xwlNyXswVYI7+Dr4iJRe8taxYP0HJ2ly0is2uJUwxASs64if 8NxWF/099aMnJRgFnz2SAQMm+IpZkD61qWg18qxNQ50v/bOXlxHSluDbt2PwYuJo v3EbuloWh/Vaftpr2hg6omvLY1EHmjVzuytxC2UyGXIb40r7JCLjlzgSKq66HDHR vdsw03s4xkKNEip/Ckx7kRlZ/cNCzv2qfQjLa8JQHT5YbyClSOifix3vjbdkNSVD U7Gb+E8jEVcAuuPNaQnzP6kFu2j9PPhmNifb0lRjxVn880D9kIEv7SSoouzzhr13 tlyRCJHhx3YUXm9bwuQUrBiuygVtYL6hLWynszm3MgqmhFrrzwlrl/rWMRDwWvGK VZby8ybGwX4lj4cLrVDpbil8YLJmjLsGhjaguxBW6ewiibL4nkw=
=B/ms
-----END PGP SIGNATURE-----

--
|_|O|_| Github: https://github.com/dpurgert
|_|_|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
|O|O|O|

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 02/02/2022 14:55, Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine.  I had an old
(ca. 2010) AMD PhenomII-based machine that ran Zoom fine.  Sure, it
couldn't do anything fancy like a virtual background or anything, but
that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based Latitude, neither of which can keep up with Discord/WebEx/Zoom.

- "[...] belongs to the entry-level graphics processor category."
- "[...] since they don’t have their own video memory, the Iris
    Xe can’t compete against the full-blown gaming-class video cards." >>> These make it sound as if buying a box with this won't help with
my issues.

No, you're reading far too much into it.  Only thing they mean is that
it's not gonna compete with the top-end monsters that nVidia and AMD put
out. Which is perfectly fine - you don't need some $1000 monster card to
join a zoom conference.

Okay. Didn't really want to spend the extra money. On the other hand,
didn't
want to buy a computer to solve a specific problem that ended up not
solving
the problem.

[...]
- "Great 1080p gaming with high visual settings"
But, is what's good for gaming automatically good for teleconferencing?

It depends on what you mean by "good for".  An nVidia 3050 is certainly
capable of it ...

But, if I understand you correctly, it's serious overkill for my purposes.

Exactly so. Modern integrated graphics are perfectly good for Zoom et
al. Don't bother with a dedicated GPU just for that.

Spend the money on RAM. 8GB minimum or 16GB for a bit of "leg room".

Is it worth twice as much as the first? Obviously, if the first one
won't do teleconferencing, its value to me is nothing. But, I don't
want to spend an extra $400 if it's not necessary.

The extra $400 is because you're getting one of those nVidia monster
cards. And yes, you're just throwing money away if all you need is
"conferencing".

I guess that I'll go ahead and buy the first one. Thanks.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

"Michael F. Stemper" <michael.stemper@gmail.com> writes:

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU utilization gets up to 300%.

Did you ever mention in the thread what it actually is in your system
that takes 300% of CPU?

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 02/02/2022 10.37, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old
(ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it
couldn't do anything fancy like a virtual background or anything, but
that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based
Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these machines have? If it's 4GiB or less, it's a lot more likely that's your problem (or rather, the heavy swapping that'd be necessary to run).

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU utilization gets up to 300%.


--
Michael F. Stemper
Isaiah 58:6-7

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 04/02/2022 08.31, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 02/02/2022 10.37, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old >>>>> (ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it
couldn't do anything fancy like a virtual background or anything, but >>>>> that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based >>>> Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these
machines have? If it's 4GiB or less, it's a lot more likely that's your
problem (or rather, the heavy swapping that'd be necessary to run).

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU
utilization gets up to 300%.

4G is pushing it with Zoom, but it "might" work if nothing else is open.

CPU spinning that hard (plus the "it works for a little bit" from one of
your other posts) sounds a lot like it's going into thermal throttling.

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

I actually capture CPU temp every minute with a cronjob. Looking at the
log from last night's Zoom meeting, it did start breaking 80C about the
time that I started having meeting content break up.

Also, yesterday morning's Discord call with my son shows CPU temp (on a different box) up into the 90s and briefly breaking 100C.

I would have thought that these were symptoms, not causes. Could you
elaborate?

Could also be software rendering for whatever reason... although usually
your DE will also cry about that.

I've never seen any complaint from that. However, during last night's Zoom meeting, Zoom was complaining that "Your high CPU activity is <something

", which I found disingenuous since Zoom was causing the high CPU

activity.

Thanks for your thoughts.


--
Michael F. Stemper
Economists have correctly predicted seven of the last three recessions.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 04/02/2022 08.04, Anssi Saari wrote:

"Michael F. Stemper" <michael.stemper@gmail.com> writes:

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU
utilization gets up to 300%.

Did you ever mention in the thread what it actually is in your system
that takes 300% of CPU?

Well, last night I was in a Zoom meeting, and the single "Zoom"
process gradually went up to over 300%, close to 350%.

WebEx is a little trickier. There, it's an aggregate of
"Web Content", "firefox", and "Xorg".

With Discord, it's three separate "Discord" processes, along
with "compiz" and "Xorg".

(In my attempts to track down the problem myself, I ran top in
batch mode a few times and wrote a program to parse the results.)

--
Michael F. Stemper
Economists have correctly predicted seven of the last three recessions.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 02/02/2022 10.37, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old
(ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it
couldn't do anything fancy like a virtual background or anything, but
that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based
Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these
machines have? If it's 4GiB or less, it's a lot more likely that's your
problem (or rather, the heavy swapping that'd be necessary to run).

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU utilization gets up to 300%.

4G is pushing it with Zoom, but it "might" work if nothing else is open.

CPU spinning that hard (plus the "it works for a little bit" from one of
your other posts) sounds a lot like it's going into thermal throttling.

Could also be software rendering for whatever reason... although usually
your DE will also cry about that.

-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE3asj+xn6fYUcweBnbWVw5UznKGAFAmH9OPsACgkQbWVw5Uzn KGBMag/7Bg//MJlsjhoNfCu/cDWbQY01tNX/EaHj+RnHFhcgyySTufC449e9LA9J s+J4M8zE0/53NR4sSnFraD/atqboZzfcyb6m6QQ3ph38A9v7UowZANVBKq9KtwMN EFZsE7XUV+r5Ah8XRmx8BDVWb5Em+VIadbqBoIn+muJfsertF6YRdZ710L7hT4QK kTvXRbAxw7p8bb2CnBQ0DbcNYdn3Gt11nmQ52alk6t//4UXP1GL1KAbER+FlsaOp OKkZ1jc57U8LTV6nioegnMN8u4jbU7wkxyrz+emDv2jcdBPLVzeUZ2MMkjBi/qGM zzyUvGpIDbGA5b/SgMKeBZdiuSl4A4Cnz9PZPBT3crcObEUa6hD2iAsqpTwJ/A9s fo96140/fxZocHK/4QwrR1DLlthXPwAL+pxaGQZpuGE6cduH/xw/bLDNGhaBl2hW QR9PpPwnIIOnqNQ+UORLvS1rgmUFbYNvN+kLaCYphrZ9HpQs0QIb18wu+ykaRtm3 QZavQLpZlumAlspsSr9SwyasJP0MdRR4xTMuD2s9wAm4WE1YtmDTtojY/3N6WNZK yS7eLC0fZdTYqnH9PhitVVQ0Oek2cvlMlYbUlgnCdsDZw+lCzWbAcpYHnCXSGC5n Unr36wS4JkKU1MPhPAUiWMbkwAz5M72B3mZUpNECCG8cQwo8Vnk=
=REnf
-----END PGP SIGNATURE-----

--
|_|O|_| Github: https://github.com/dpurgert
|_|_|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
|O|O|O|

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 02/02/2022 10.37, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine. I had an old >>>>>> (ca. 2010) AMD PhenomII-based machine that ran Zoom fine. Sure, it >>>>>> couldn't do anything fancy like a virtual background or anything, but >>>>>> that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based >>>>> Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these >>>> machines have? If it's 4GiB or less, it's a lot more likely that's your >>>> problem (or rather, the heavy swapping that'd be necessary to run).

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU
utilization gets up to 300%.

4G is pushing it with Zoom, but it "might" work if nothing else is open.

CPU spinning that hard (plus the "it works for a little bit" from one of
your other posts) sounds a lot like it's going into thermal throttling.

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it
can cool down.

For the sake of discussion, let's say your CPU normally operates at 2
GHz. Under a normal system load, maybe the CPU is 10% loaded and
operating at a temperature of 50C.

You fire up Zoom (etc) and now the CPU is at 80% load, and starts
generating considerably more heat (which gets dumped into the heatsink
at first, which gets cooled by air blowing across it).

Due to dust (causes poor heat transfer to the air / poor airflow), or insufficient thermal grease between the heatsink and CPU, or a bad fan,
your heatsink is unable to adequately cool things down, so what starts happening is the whole shebang starts getting hotter and hotter.

At some point, the CPU die gets hot enough that it hits a thermal
cutout, and throttles itself down from 2.00 GHz to say 500 MHz. This significantly slower clockrate slows the generation of heat within the
CPU and allows your thermal management system (heatsink & fans) to try
catching up.

Thing is ... now that the processor is "only 500 MHz", it's far too slow
to do things like run Zoom -- and your load skyrockets from 0.8 to 3.2
(or higher), since applications now need to wait 4x as long for the
processor to do its thing.


Cheap and super-easy fix -> blow out the heatsink fins (BACKWARDS) with compressed air (you can obtain at walmart, etc. in the electronics
dept). That is, point the nozzle of the air can into the fan exhaust so
that any caked on dust is dislodged and blown back out the intake.

More involved cheap fix -> open the case, remove the fan, blow more air
through heatsink fins. Optionally, replace the fan if it doesn't spin
very easily (NOTE modern brushless motors "cog" a little bit as the
permanent magnets in the rotor find the ferrite cores of the
electromagnets in the stator).

Involved and inexpensive (though not "cheap") fix -> in addition to the
above, take out the heatsink, clean off all the old thermal grease, and
apply a new coating, then reinstall. If you've never done this before,
I wouldn't do it on a machine that I _NEEDED_ for the first time (it can
be fiddly). That 2010 model you have (or other "backup" PC that you
don't particularly _NEED_ to continue working) would be the safest one
to get a feeling on. Basically if you mess up, the CPU overheats faster
than if you'd done nothing at all.


-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE3asj+xn6fYUcweBnbWVw5UznKGAFAmH9U9EACgkQbWVw5Uzn KGBwag//WkeaU74HcziMN2M7EV2fcrdxTh8Vku6Nb0kEE535S5hew7yYcuKil5HU BXnrHy5V5RweTGl+RYu+GLqBnTmspVX1XtfyoHWcrKAlL74X1fpHkRf2dZxyNLsw WwUQ2Wt+ZixFyIijk07N00qU9qhOvkUniqBJEVFUxAJcUyYCil4UkEsmRWZ9nlot gccQfcTBdQrZivcnV1Bfb2Unm4/RnwMoj2Bjzf7D6btSq8yfhNtDQEfrrFm4YTkP 9vWHAAwc+bDtWG+i7sY1ylXcKMTqKDC8b4U4ZfFEaCqvR50JrsoGmd10hNT9D5X7 ubSllIuRxeUopAMSBKhuyziVtTS6zzcbmLoO9vfIAKNL7w8z28Mh1OsHlBnZJEGq Agq0k4HYBolHkXXQbz+AAhAqx4Iw0WbF3yihbAD3vfcg0JCGj7Z0OVazlB6afodl JUe27qIhEP3UYZ4IdT51RAPrnfpZk43rPB8r+y0aYMpTdJExer4nK2Eg7c/IezOE 5ff2dp82pJ3NqoncwfGZfu5Tx2WEsutwElcCOm2n5lw4IW9n46DXLZWDQxoHrAN0 +JZ+BAONEIBPc/cMah5lhfrEarwZADzyLdbb0m5XthC5IBuJNsQuBxFaN6Sjn1Bg xPsYFXtUwLGxBNcIJLj8aNcftMuY6NRuL5cifiTg5UskFi6lneM=
=0O6n
-----END PGP SIGNATURE-----

--
|_|O|_| Github: https://github.com/dpurgert
|_|_|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
|O|O|O|

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 04/02/2022 10.26, Dan Purgert wrote:

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

Michael F. Stemper wrote:

On 02/02/2022 10.37, Dan Purgert wrote:

Michael F. Stemper wrote:

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based >>>>>> Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these >>>>> machines have? If it's 4GiB or less, it's a lot more likely that's your >>>>> problem (or rather, the heavy swapping that'd be necessary to run).

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU
utilization gets up to 300%.

4G is pushing it with Zoom, but it "might" work if nothing else is open. >>>
CPU spinning that hard (plus the "it works for a little bit" from one of >>> your other posts) sounds a lot like it's going into thermal throttling.

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it
can cool down.

For the sake of discussion, let's say your CPU normally operates at 2
GHz. Under a normal system load, maybe the CPU is 10% loaded and
operating at a temperature of 50C.

You fire up Zoom (etc) and now the CPU is at 80% load, and starts
generating considerably more heat (which gets dumped into the heatsink
at first, which gets cooled by air blowing across it).

Due to dust (causes poor heat transfer to the air / poor airflow), or insufficient thermal grease between the heatsink and CPU, or a bad fan,
your heatsink is unable to adequately cool things down, so what starts happening is the whole shebang starts getting hotter and hotter.

At some point, the CPU die gets hot enough that it hits a thermal
cutout, and throttles itself down from 2.00 GHz to say 500 MHz. This significantly slower clockrate slows the generation of heat within the
CPU and allows your thermal management system (heatsink & fans) to try catching up.

Thing is ... now that the processor is "only 500 MHz", it's far too slow
to do things like run Zoom -- and your load skyrockets from 0.8 to 3.2
(or higher), since applications now need to wait 4x as long for the
processor to do its thing.

Ah, the CPU utilization isn't based on nameplate rating, but on its
current capability. That makes a lot of sense. Thanks.

I will try the compressed air things.

Involved and inexpensive (though not "cheap") fix -> in addition to the above, take out the heatsink, clean off all the old thermal grease, and
apply a new coating, then reinstall. If you've never done this before,
I wouldn't do it on a machine that I _NEEDED_ for the first time (it can
be fiddly). That 2010 model you have (or other "backup" PC that you
don't particularly _NEED_ to continue working) would be the safest one
to get a feeling on. Basically if you mess up, the CPU overheats faster
than if you'd done nothing at all.

I have another one that's newer, that I didn't mention because it's on
its second bad keyboard. (I'm not gonna get an external one because
space.) If I feel really adventurous, I'll try the above on it.


--
Michael F. Stemper
If it isn't running programs and it isn't fusing atoms, it's just bending space.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2/4/22 09:15, Michael F. Stemper wrote:

On 04/02/2022 10.26, Dan Purgert wrote:

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

Michael F. Stemper wrote:

On 02/02/2022 10.37, Dan Purgert wrote:

Michael F. Stemper wrote:

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel
i5-based
Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do
these
machines have? If it's 4GiB or less, it's a lot more likely that's >>>>>> your
problem (or rather, the heavy swapping that'd be necessary to run). >>>>>

The Inspiron has 3 GiB and the Latitude has 4. However, top says that >>>>> memory usage (on either) is only in the 15%-20% range, even when CPU >>>>> utilization gets up to 300%.

4G is pushing it with Zoom, but it "might" work if nothing else is
open.

CPU spinning that hard (plus the "it works for a little bit" from
one of
your other posts) sounds a lot like it's going into thermal throttling. >>>

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it
can cool down.

For the sake of discussion, let's say your CPU normally operates at 2
GHz. Under a normal system load, maybe the CPU is 10% loaded and
operating at a temperature of 50C.

You fire up Zoom (etc) and now the CPU is at 80% load, and starts
generating considerably more heat (which gets dumped into the heatsink
at first, which gets cooled by air blowing across it).

Due to dust (causes poor heat transfer to the air / poor airflow), or
insufficient thermal grease between the heatsink and CPU,  or a bad fan,
your heatsink is unable to adequately cool things down,  so what starts
happening is the whole shebang starts getting hotter and hotter.

At some point, the CPU die gets hot enough that it hits a thermal
cutout, and throttles itself down from 2.00 GHz to say 500 MHz.  This
significantly slower clockrate slows the generation of heat within the
CPU and allows your thermal management system (heatsink & fans) to try
catching up.

Thing is ... now that the processor is "only 500 MHz", it's far too slow
to do things like run Zoom -- and your load skyrockets from 0.8 to 3.2
(or higher), since applications now need to wait 4x as long for the
processor to do its thing.

Ah, the CPU utilization isn't based on nameplate rating, but on its
current capability. That makes a lot of sense. Thanks.

I will try the compressed air things.

Involved and inexpensive (though not "cheap") fix -> in addition to the
above, take out the heatsink, clean off all the old thermal grease, and
apply a new coating, then reinstall.  If you've never done this before,
I wouldn't do it on a machine that I _NEEDED_ for the first time (it can
be fiddly).  That 2010 model you have (or other "backup" PC that you
don't particularly _NEED_ to continue working) would be the safest one
to get a feeling on.  Basically if you mess up, the CPU overheats faster
than if you'd done nothing at all.

I have another one that's newer, that I didn't mention because it's on
its second bad keyboard. (I'm not gonna get an external one because
space.) If I feel really adventurous, I'll try the above on it.

Get a regular keyboard if that backup has a USB or other keyboard port.
I use a external keyboard on all my regular machines
as repeated blows from fingers on a built-in keyboard will reveal
weak points in construction of the devices. Use a passive USB extender
amd an easel with cooling built in if you can. I am sure that
external keyboards increase reliability when you must use the built-in keyboard.

Just a tiny bit of advice from a user who has lost keytops on
my earlier machines.

bliss - brought to you by the power and ease of PCLinuxOS
and a minor case of hypergraphia

--
bliss dash SF 4 ever at dslextreme dot com

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 04/02/2022 12.26, Bobbie Sellers wrote:

On 2/4/22 09:15, Michael F. Stemper wrote:

I have another one that's newer, that I didn't mention because it's on
its second bad keyboard. (I'm not gonna get an external one because
space.) If I feel really adventurous, I'll try the above on it.

Get a regular keyboard if that backup has a USB or other keyboard port.

As I said, I'm not going to do that, due to space constraints.

--
Michael F. Stemper
87.3% of all statistics are made up by the person giving them.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2/4/2022 10:02 AM, Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 02/02/2022 10.37, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 31/01/2022 04.42, Dan Purgert wrote:

Michael F. Stemper wrote:

On 24/01/2022 14.26, Paul wrote:

On 1/24/2022 10:49 AM, Michael F. Stemper wrote:

I think that your main point was that I should read reviews of
the graphics card before buying something.

Honestly, anything "new" will do teleconferencing fine.  I had an old >>>>>> (ca. 2010) AMD PhenomII-based machine that ran Zoom fine.  Sure, it >>>>>> couldn't do anything fancy like a virtual background or anything, but >>>>>> that wasn't super important.

Well, I have 2010 Pentium Dual-based Inspiron and a 2015 Intel i5-based >>>>> Latitude, neither of which can keep up with Discord/WebEx/Zoom.

On the off chance I missed the comments before - how much RAM do these >>>> machines have? If it's 4GiB or less, it's a lot more likely that's your >>>> problem (or rather, the heavy swapping that'd be necessary to run).

The Inspiron has 3 GiB and the Latitude has 4. However, top says that
memory usage (on either) is only in the 15%-20% range, even when CPU
utilization gets up to 300%.

4G is pushing it with Zoom, but it "might" work if nothing else is open.

CPU spinning that hard (plus the "it works for a little bit" from one of
your other posts) sounds a lot like it's going into thermal throttling.

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

I actually capture CPU temp every minute with a cronjob. Looking at the
log from last night's Zoom meeting, it did start breaking 80C about the
time that I started having meeting content break up.

Also, yesterday morning's Discord call with my son shows CPU temp (on a different box) up into the 90s and briefly breaking 100C.

I would have thought that these were symptoms, not causes. Could you elaborate?

Could also be software rendering for whatever reason... although usually
your DE will also cry about that.

I've never seen any complaint from that. However, during last night's Zoom meeting, Zoom was complaining that "Your high CPU activity is <something bad>", which I found disingenuous since Zoom was causing the high CPU activity.

Thanks for your thoughts.

sudo apt install inxi

inxi -G # gives driver details

sudo apt install mesa-utils

vblank=0 glxgears

*******

bullwinkle@Roomba:~$ inxi -G
Graphics:
Device-1: AMD driver: amdgpu v: kernel
Display: x11 server: X.Org 1.20.13 driver: amdgpu,ati <=== seems accelerated when I move a window
unloaded: fbdev,modesetting,vesa resolution: 1440x900~60Hz
OpenGL: renderer: AMD RENOIR (DRM 3.41.0 5.13.0-27-generic LLVM 12.0.0)
v: 4.6 Mesa 21.0.3

bullwinkle@Roomba:~$ vblank_mode=0 glxgears
ATTENTION: default value of option vblank_mode overridden by environment. 101083 frames in 5.0 seconds = 20216.369 FPS
100957 frames in 5.0 seconds = 20191.338 FPS <=== this means I probably have the right
103397 frames in 5.0 seconds = 20679.207 FPS OpenGL acceleration going... That's my
^C AMD processor GPU, so a bit weak.
bullwinkle@Roomba:~$

*******

My driver is "amdgpu", not just a "vesa".

I also don't get a warning in LinuxMint that my screen
rendering is unaccelerated.

The "glxgears" test, actually evaluates that OpenGL acceleration
is present. A value of 60FPS, means the "vblank" thing
did not override the settings. A value of 3000FPS would be
"weak" OpenGL. A value of 20000-25000 is pretty good.

The video decoder suite, hardware accelerated, is
more difficult to evaluate. But, you can try the client
settings box for this. It does not look, in this support
page, like the client lists the "flavors" of acceleration.

https://support.zoom.us/hc/en-us/articles/360037870291-Managing-advanced-video-settings

"Use hardware acceleration for receiving video (macOS, Linux):

Utilizes hardware resources to improve rendering of received video feeds;
if your system does not have the required hardware resources, this can
make the image worse."

You can try this (for Intel/ATI, example is ATI):

sudo apt install vainfo

vainfo

bullwinkle@Roomba:~$ vainfo
libva info: VA-API version 1.7.0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/radeonsi_drv_video.so libva info: Found init function __vaDriverInit_1_7
libva info: va_openDriver() returns 0
vainfo: VA-API version: 1.7 (libva 2.6.0)
vainfo: Driver version: Mesa Gallium driver 21.0.3 for AMD RENOIR (DRM 3.41.0, 5.13.0-27-generic, LLVM 12.0.0)
vainfo: Supported profile and entrypoints
VAProfileMPEG2Simple : VAEntrypointVLD
VAProfileMPEG2Main : VAEntrypointVLD

VAProfileVC1Simple : VAEntrypointVLD
VAProfileVC1Main : VAEntrypointVLD
VAProfileVC1Advanced : VAEntrypointVLD

VAProfileH264ConstrainedBaseline: VAEntrypointVLD
VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointEncSlice
VAProfileH264High : VAEntrypointVLD \___ useful for conferences ?
VAProfileH264High : VAEntrypointEncSlice /

VAProfileHEVCMain : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointEncSlice
VAProfileHEVCMain10 : VAEntrypointVLD
VAProfileHEVCMain10 : VAEntrypointEncSlice

VAProfileJPEGBaseline : VAEntrypointVLD

VAProfileVP9Profile0 : VAEntrypointVLD
VAProfileVP9Profile2 : VAEntrypointVLD

VAProfileNone : VAEntrypointVideoProc

Some background info. You can also see some conceptual overlap,
as companies promote their own scheme.

https://en.wikipedia.org/wiki/Video_Acceleration_API <=== Intel https://en.wikipedia.org/wiki/VDPAU <=== some others

For VDPAU, I found this. It pretty well requires that you
have an NVidia GPU and have the NVidia driver loaded. It's
dynamically linked, and I don't know if it would tolerate
static linking and portability all that well. It needs
things like build-essential, cmake, qmake, libvdpau1, libvdpau-dev, mesa-libvdpau, (nvidia-driver from driver-manager in LinuxMint).

https://github.com/robertmassaioli/qvdpautest

This is a test run. NVidia decoding acceleration.

*******

bullwinkle@Bungalow:~/Downloads/qvdpautest-master/data$ ./qvdpautest
qvdpautest 0.5.2++
Intel(R) Core(TM) i7-4930K CPU @ 3.40GHz
Unknown apple GPU [GTX1080 on a PC...]

VDPAU API version : 1
VDPAU implementation : NVIDIA VDPAU Driver Shared Library 495.46 Wed Oct 27 16:20:01 UTC 2021

SURFACE GET BITS: 1535.05 M/s
SURFACE PUT BITS: 1829.62 M/s

MPEG DECODING (1920x1080): 561 frames/s
MPEG DECODING (1280x720): 1256 frames/s
H264 DECODING (1920x1080): 559 frames/s \___ useful for conferences ?
H264 DECODING (1280x720): 909 frames/s /
VC1 DECODING (1440x1080): 931 frames/s
MPEG4 DECODING (1920x1080): 717 frames/s

MIXER WEAVE (1920x1080): 13580 frames/s
MIXER BOB (1920x1080): 22683 fields/s
MIXER TEMPORAL (1920x1080): 6875 fields/s
MIXER TEMPORAL + IVTC (1920x1080): 4020 fields/s
MIXER TEMPORAL + SKIP_CHROMA (1920x1080): 9028 fields/s
MIXER TEMPORAL_SPATIAL (1920x1080): 4376 fields/s
MIXER TEMPORAL_SPATIAL + IVTC (1920x1080): 2806 fields/s
MIXER TEMPORAL_SPATIAL + SKIP_CHROMA (1920x1080): 5148 fields/s
MIXER TEMPORAL_SPATIAL (720x576 video to 1920x1080 display): 10337 fields/s MIXER TEMPORAL_SPATIAL + HQSCALING (720x576 video to 1920x1080 display): 7881 fields/s

MULTITHREADED MPEG DECODING (1920x1080): 621 frames/s
MULTITHREADED MIXER TEMPORAL (1920x1080): 5853 fields/s

QClipboard: Unable to receive an event from the clipboard manager in a reasonable time
*******

A potential reason Zoom could start to rail, could be
memory fragmentation the longer it runs. Sometimes, bad
memory management results in increased resource usage
as time passes. That doesn't mean you are short of memory,
it just means the allocation scheme and the garbage
collection, are not optimal in the coding. Or, if
the operations are browser-accelerated (they use a browser
engine like XUL.so and not the browser itself), it is the browser
engine which is sub-optimal.

You might in such cases, still need to review the browser
acceleration settings (like in about:config), to see if
hardware acceleration is still turned on in there.

Paul

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 04/02/2022 10.26, Dan Purgert wrote:

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

CPU spinning that hard (plus the "it works for a little bit" from one of >>> your other posts) sounds a lot like it's going into thermal throttling.

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it
can cool down.

For the sake of discussion, let's say your CPU normally operates at 2
GHz. Under a normal system load, maybe the CPU is 10% loaded and
operating at a temperature of 50C.

You fire up Zoom (etc) and now the CPU is at 80% load, and starts
generating considerably more heat (which gets dumped into the heatsink
at first, which gets cooled by air blowing across it).

Due to dust (causes poor heat transfer to the air / poor airflow), or insufficient thermal grease between the heatsink and CPU, or a bad fan,
your heatsink is unable to adequately cool things down, so what starts happening is the whole shebang starts getting hotter and hotter.

At some point, the CPU die gets hot enough that it hits a thermal
cutout, and throttles itself down from 2.00 GHz to say 500 MHz. This significantly slower clockrate slows the generation of heat within the
CPU and allows your thermal management system (heatsink & fans) to try catching up.

Thing is ... now that the processor is "only 500 MHz", it's far too slow
to do things like run Zoom -- and your load skyrockets from 0.8 to 3.2
(or higher), since applications now need to wait 4x as long for the
processor to do its thing.

Cheap and super-easy fix -> blow out the heatsink fins (BACKWARDS) with compressed air (you can obtain at walmart, etc. in the electronics
dept). That is, point the nozzle of the air can into the fan exhaust so
that any caked on dust is dislodged and blown back out the intake.

More involved cheap fix -> open the case, remove the fan, blow more air through heatsink fins. Optionally, replace the fan if it doesn't spin
very easily (NOTE modern brushless motors "cog" a little bit as the
permanent magnets in the rotor find the ferrite cores of the
electromagnets in the stator).

I opened up one of them and pulled out a dust bunny bigger than I ever
expect to see under a sofa.

This evening, I had a zoom meeting that lasted an hour and a quarter.
This box had no problems keeping up; CPU utilization seldom got over
120%; CPU temperature hit 72 C at its highest.

Tomorrow morning's discord call will be another data point, but it looks
as if we have a winner. Thanks much!

--
Michael F. Stemper
87.3% of all statistics are made up by the person giving them.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 04/02/2022 10.26, Dan Purgert wrote:

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

CPU spinning that hard (plus the "it works for a little bit" from one of >>>> your other posts) sounds a lot like it's going into thermal throttling. >>>

I hadn't heard of that before, but it sounds like something that would
reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in
would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it
can cool down.
[...]

I opened up one of them and pulled out a dust bunny bigger than I ever
expect to see under a sofa.

This evening, I had a zoom meeting that lasted an hour and a quarter.
This box had no problems keeping up; CPU utilization seldom got over
120%; CPU temperature hit 72 C at its highest.

Is that "utilization" as from looking at the oad average? If so, AND
the machine is multi-core; "1.2" load average is 60% or lower loading
per core.

And glad to hear it turned out to be a simple fix :)

-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE3asj+xn6fYUcweBnbWVw5UznKGAFAmIE5tQACgkQbWVw5Uzn KGAhhQ/+Jif0LF+K9FCFahIzpE4BnNnSsDQXGIw8f9uED4Kv32KRyUWkn9pWJYuz 760eWXOxtWwt9IO4V2PfUkgiwo2UMPjG9Qxvfa388BG3TEevp68gA5novvJK+v38 npxtQNqAP1i2ZCIVbTpUDEPnUJ8kxX/oqOwiMgCWHZZsBuM8rDnmJuQmDIasWzhx KBXk2uz+EQaPiVj1Rvodth/FQRGEhaTC1oB6QDy3NuiJlmXhDmdLlJv5Ii2DDaOg ABKjvQhfhocYyulQtCelGBBNh+nOhxeZTu7YGOTbxyX4LQt2GUftZ1XyMVd+T5w3 64uhCvCQ0xwHxevFMNIdz7wzZ7T7H9XaMNkGms8AqETYF5bLighgKKgDuPyJ+GmX CuJRPZmNUvHFUT0X0ahjNvb3N1lOfd9ePOHQ6fJ5tlLQGwUK8r72XpIvNlDmcdr8 EQCu1H4MFlnRDEqpwSX99/gKWhCAzAmcST6OUHCjVGPALJtSJ7L0y6vWjTAp742u kP6jvzZWFARPMTlf7aNndGMWKw2DluRDgqK6qARz7U3PoANYhjagXsBd71F7V4KY jKSY7Q0uq6CTKobc6arD7v+Q4MhkMBDUEbYj5vvaAg2vF96maJAuHaxc7sSYx3Hm 0wUzZQgauV+RlyK7UE+EpnoGCSf+Z/mQaRwau3DS01KwNYNAv+o=
=SIr5
-----END PGP SIGNATURE-----

--
|_|O|_| Github: https://github.com/dpurgert
|_|_|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
|O|O|O|

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 10/02/2022 04.19, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 04/02/2022 10.26, Dan Purgert wrote:

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

CPU spinning that hard (plus the "it works for a little bit" from one of >>>>> your other posts) sounds a lot like it's going into thermal throttling. >>>>

I hadn't heard of that before, but it sounds like something that would >>>> reduce CPU utilization, not increase it. What am I misunderstanding?
I certainly agree that the time delay before performance issues kick in >>>> would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it >>> can cool down.
[...]

I opened up one of them and pulled out a dust bunny bigger than I ever
expect to see under a sofa.

This evening, I had a zoom meeting that lasted an hour and a quarter.
This box had no problems keeping up; CPU utilization seldom got over
120%; CPU temperature hit 72 C at its highest.

Is that "utilization" as from looking at the oad average? If so, AND
the machine is multi-core; "1.2" load average is 60% or lower loading
per core.

It's the value reported by top. For comparison, my previous zoom (from
which I had to bail) was running 300% to 350%.

I have no idea what "oad" is:

user@host$ man oad
No manual entry for oad
user@host$

And glad to hear it turned out to be a simple fix :)

As am I. Maybe when I discord with my son (half an hour from now), we'll
be able to see each other, and converse in real time. :->

--
Michael F. Stemper
Psalm 94:3-6

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 10/02/2022 04.19, Dan Purgert wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Michael F. Stemper wrote:

On 04/02/2022 10.26, Dan Purgert wrote:

Michael F. Stemper wrote:

On 04/02/2022 08.31, Dan Purgert wrote:

CPU spinning that hard (plus the "it works for a little bit" from one of >>>>>> your other posts) sounds a lot like it's going into thermal throttling. >>>>>

I hadn't heard of that before, but it sounds like something that would >>>>> reduce CPU utilization, not increase it. What am I misunderstanding? >>>>> I certainly agree that the time delay before performance issues kick in >>>>> would align with it being temperature-related.

Other way around -- CPU gets hot, CPU stops operating "fast", so that it >>>> can cool down.
[...]

I opened up one of them and pulled out a dust bunny bigger than I ever
expect to see under a sofa.

This evening, I had a zoom meeting that lasted an hour and a quarter.
This box had no problems keeping up; CPU utilization seldom got over
120%; CPU temperature hit 72 C at its highest.

Is that "utilization" as from looking at the oad average? If so, AND
the machine is multi-core; "1.2" load average is 60% or lower loading
per core.

It's the value reported by top. For comparison, my previous zoom (from
which I had to bail) was running 300% to 350%.

I have no idea what "oad" is:

oops, typo. It's "load average", you can see it on the first line of
output when running 'top':

09:35:47 up 95 days, 21:01, 1 user, load average: 0.07, 0.03, 0.01

show a running 1, 5, 15 minute "average load" of the system; although
gets thrown off when you have a multi-core system.

In the past, a load average of "1.00" meant your CPU was running at
capacity. These days, the whole number would need to equal the number
of cores/threads you have (so a 4-core system could go as high as "4.00"
before being fully loaded down, and a load average of "8" would mean
you've got twice as much stuff scheduled as the CPU can handle).

I find it to be a little easier to follow than the "%CPU(s)" line a bit
below that (as that jumps around with frequency a little too much -- or
used to anyway).

-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE3asj+xn6fYUcweBnbWVw5UznKGAFAmIFJOsACgkQbWVw5Uzn KGCpzBAAlGCMddUSwoSlMLAmO/Sx3p+soA4Wz+bACWLTbi7a/XMjJS7jvaWUFxkJ jbAr90vdM4+QvdB/EB9GxiLQqyxuNbODo6v/2nMCz/XXN8K5z/q/OPblJNvMRM+m wx2GDOzvwxbFDHNw/eEBmAe6lsdZUWOzVn3+9hKDYUwoDjVQihUqmyRl8rByrKMR ygn/6xbHwVZakE4pumdDq46H5SbXktq/ncbY/GypejdtMBbO8Uh1kdV2bGlS6r1i ToWVMctGPCEYdglLJ7Kw9jFpSjJ4r8oPdATtgnU+0IixoIz5qLMK9gBTTn556O7L yZ7X2vCbNf0NDw2TACfGO/4BSbUnSt9VfJ6cP8RVnSKDVEa2NRskAMV65B7+0aAB 95SsZCRxVP1sn3kKqNQ3YoPB10Lmka/AQRF7Eh4oFRhFM3QVhFWVXzuFjwj8/J0z 0JgTcDdhowRLFl9/zj6dxBQFH+C1+dTvGly7D467ivUWAt1WCAskLpPP99Ws798Q /+AkMmsnyT+DHd9KYU2WjswvE0SoNNyyE3SPNHWx29iv0p6pBgVOZtEONWHIsBAZ 3sAq21jczYzIEN0IHNxLXo1nMRlOELxIYoPMytiIMhFWLI+z9CHhFUb/qnTNxFUK WbvLm3bzaqz/sDR7bsmgpcKgeai17VlgNtiKGeY06sW1CvKzbwo=
=90J6
-----END PGP SIGNATURE-----

--
|_|O|_| Github: https://github.com/dpurgert
|_|_|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860
|O|O|O|

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Dan Purgert <dan@djph.net> writes:

It's "load average", you can see it on the first line of
output when running 'top':

09:35:47 up 95 days, 21:01, 1 user, load average: 0.07, 0.03, 0.01

show a running 1, 5, 15 minute "average load" of the system; although
gets thrown off when you have a multi-core system.

In the past, a load average of "1.00" meant your CPU was running at
capacity. These days, the whole number would need to equal the number
of cores/threads you have (so a 4-core system could go as high as "4.00" >before being fully loaded down, and a load average of "8" would mean
you've got twice as much stuff scheduled as the CPU can handle).

I find it to be a little easier to follow than the "%CPU(s)" line a bit
below that (as that jumps around with frequency a little too much -- or
used to anyway).

Yes. However, if you type "1" in top, you get the %CPU(s) line split
into one line for each core/thread, e.g.:

%Cpu0 : 1.0 us, 0.7 sy, 0.0 ni, 98.3 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st %Cpu1 : 1.7 us, 0.3 sy, 0.0 ni, 97.7 id, 0.3 wa, 0.0 hi, 0.0 si, 0.0 st %Cpu2 : 0.3 us, 0.7 sy, 0.0 ni, 99.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st %Cpu3 : 0.7 us, 0.3 sy, 0.0 ni, 98.7 id, 0.3 wa, 0.0 hi, 0.0 si, 0.0 st

us(er) mode
sy(stem) mode
ni(ce) (low priority)
id(le): the core does not have things to do during this time
wa(it): the core waits for some quick I/O completion
hi: hardware interrupts
si: software interrupts
st(olen) by hypervisor

And if you then type "t", you get an alternate display:

%Cpu0 : 1.0/0.3 1[| ] %Cpu1 : 1.0/0.3 1[| ] %Cpu2 : 1.7/0.0 2[| ] %Cpu3 : 1.3/0.3 2[| ]

user/system total histogram

If total of one core is 100% or close to it the application running on
the core is probably limited by CPU performance.

- anton
--
M. Anton Ertl Some things have to be seen to be believed anton@mips.complang.tuwien.ac.at Most things have to be believed to be seen http://www.complang.tuwien.ac.at/anton/home.html

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 10/02/2022 08.44, Dan Purgert wrote:

Michael F. Stemper wrote:

On 10/02/2022 04.19, Dan Purgert wrote:

Michael F. Stemper wrote:

This evening, I had a zoom meeting that lasted an hour and a quarter.
This box had no problems keeping up; CPU utilization seldom got over
120%; CPU temperature hit 72 C at its highest.

Is that "utilization" as from looking at the oad average? If so, AND
the machine is multi-core; "1.2" load average is 60% or lower loading
per core.

It's the value reported by top. For comparison, my previous zoom (from
which I had to bail) was running 300% to 350%.

I have no idea what "oad" is:

oops, typo. It's "load average", you can see it on the first line of
output when running 'top':

When I reopened your post, I immediately saw that "oad" was obviously
a typo for "load". My guess is that my blood caffeine level was too
low earlier.

09:35:47 up 95 days, 21:01, 1 user, load average: 0.07, 0.03, 0.01

show a running 1, 5, 15 minute "average load" of the system; although
gets thrown off when you have a multi-core system.

Cool! I never knew what those numbers meant.

In the past, a load average of "1.00" meant your CPU was running at
capacity. These days, the whole number would need to equal the number
of cores/threads you have (so a 4-core system could go as high as "4.00" before being fully loaded down, and a load average of "8" would mean
you've got twice as much stuff scheduled as the CPU can handle).

I find it to be a little easier to follow than the "%CPU(s)" line a bit
below that (as that jumps around with frequency a little too much -- or
used to anyway).

For testing purposes, I've been running top with -d60 in batch mode and re-directing its output to a file. Then, after the fact, I have a program
that grabs and totals the usage for all of the processes specified on the command line as a regex. That way, I can see, e.g., the aggregate usage
from "discord|compiz|Xorg", which separates stuff attributable to the teleconference from stuff not related to it, such as ibus-daemon.

As it has turned out, it was actually the total usage that was causing
the problem, but that's not what I thought going in.

And this morning's discord with my son went off with nary a hitch, so everything is golden.

Thanks for your help and to all the others who pitched in.

--
Michael F. Stemper
The name of the story is "A Sound of Thunder".
It was written by Ray Bradbury. You're welcome.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 10/02/2022 11.26, Anton Ertl wrote:

Dan Purgert <dan@djph.net> writes:

It's "load average", you can see it on the first line of
output when running 'top':

I find it to be a little easier to follow than the "%CPU(s)" line a bit
below that (as that jumps around with frequency a little too much -- or
used to anyway).

Yes. However, if you type "1" in top, you get the %CPU(s) line split
into one line for each core/thread, e.g.:

%Cpu0 : 1.0 us, 0.7 sy, 0.0 ni, 98.3 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu1 : 1.7 us, 0.3 sy, 0.0 ni, 97.7 id, 0.3 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu2 : 0.3 us, 0.7 sy, 0.0 ni, 99.0 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
%Cpu3 : 0.7 us, 0.3 sy, 0.0 ni, 98.7 id, 0.3 wa, 0.0 hi, 0.0 si, 0.0 st

That is cool! If I run a compute-intensive program, I can watch it
bounce around from CPU to CPU.

Thanks for the tip.

--
Michael F. Stemper
Galatians 3:28

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)