On 2023-11-16 23:29, RabidPedagog wrote:
> On 2023-11-16 4:19 p.m., Carlos E. R. wrote:
>> On 2023-11-16 22:05, RabidPedagog wrote:
>>> On 2023-11-16 11:12 a.m., SBIANCO wrote:
>>>> Il 16/11/2023 16:46, RabidPedagog ha scritto:
>>>>
>>>>>
>>>>> Are you making an affirmation or an interrogation?
>>>>>
>>>>
>>>> I'm asking:
>>>> why my pc sometimes goes slow?
>>>
>>> Because you're opening an inordinate amount of tabs on four browsers
>>> for some ridiculous reason and expecting things to perform well.
>>
>> No, that on itself doesn't cause sluggishness. There are more factors
>> that have to concur.
>
> Well, it's rather simple. If SBIANCO isn't using a browser that blocks
> ads or JavaScript, every last one of those tabs is loading garbage
> regardless of the fact that he might not be looking at them. That's
> going to take a few CPU cycles no matter what he does.

It doesn't work like that, at least in Firefox on Linux (we don't know
yet what he is using).

Firefox doesn't run or load all tabs from the previous session, only the
one that has the focus on each window, or those you click on. The rest
are blank, and when clicking on them they reload from the cache.
Unfortunately, the only method to stop "running" tabs from running is to
close them or restart FF.

There is a setting for this, I think.

--
Cheers,
Carlos E.R.

On 2023-11-16 5:40 p.m., Carlos E. R. wrote:
> On 2023-11-16 23:29, RabidPedagog wrote:
>> On 2023-11-16 4:19 p.m., Carlos E. R. wrote:
>>> On 2023-11-16 22:05, RabidPedagog wrote:
>>>> On 2023-11-16 11:12 a.m., SBIANCO wrote:
>>>>> Il 16/11/2023 16:46, RabidPedagog ha scritto:
>>>>>
>>>>>>
>>>>>> Are you making an affirmation or an interrogation?
>>>>>>
>>>>>
>>>>> I'm asking:
>>>>> why my pc sometimes goes slow?
>>>>
>>>> Because you're opening an inordinate amount of tabs on four browsers
>>>> for some ridiculous reason and expecting things to perform well.
>>>
>>> No, that on itself doesn't cause sluggishness. There are more factors
>>> that have to concur.
>>
>> Well, it's rather simple. If SBIANCO isn't using a browser that blocks
>> ads or JavaScript, every last one of those tabs is loading garbage
>> regardless of the fact that he might not be looking at them. That's
>> going to take a few CPU cycles no matter what he does.
>
> It doesn't work like that, at least in Firefox on Linux (we don't know
> yet what he is using).
>
> Firefox doesn't run or load all tabs from the previous session, only the
> one that has the focus on each window, or those you click on. The rest
> are blank, and when clicking on them they reload from the cache.
> Unfortunately, the only method to stop "running" tabs from running is to
> close them or restart FF.
>
> There is a setting for this, I think.

As far as I know, those other tabs only stop all of their processes if
they automatically go to sleep, which is a rather new function for
browsers. Otherwise, they're all essentially active. If Firefox puts
them to sleep by default, and since a long time, then that's great.

--
RabidPedagog
TG: @RabidPedagog
Mark 13:6

On 11/15/23 20:12, rbowman wrote:
> On Wed, 15 Nov 2023 10:47:48 -0600, candycanearter07 wrote:
>
>> You *gave away all your machines*? Why?
>
> When I left New Hampshire in '88 if it didn't fit in the pickup it wasn't
> going. A teacher was trying to set up a computer course so I gave him the
> machines, development boards, manuals, general comp sci books, and so
> forth.

Oh, that was nice of you!
--
user <candycane> is generated from /dev/urandom

On 11/16/23 10:01, Mark Lloyd wrote:
> On 11/15/23 17:09, Paul wrote:
>> On 11/14/2023 11:11 PM, Relf wrote:
>>> David_LaRue:
>>>> Addng RAM will help any platform.
>>>> My 64GB for Win 7 Pro does all I need and its over 10 years old.
>>>> I'm not a gamer though.
>>>
>>> What are you doing that requires 64 GB of RAM ?
>>>
>>
>> RAM is never wasted. You can always put it to use.
>
> Its wasted if you use a program that frees it, although I haven't heard
> of anyone done that for years.
>
> [snip]
>

Frees it? Doesn't that make ram available for use?
--
user <candycane> is generated from /dev/urandom

On 2023-11-17 00:01, RabidPedagog wrote:
> On 2023-11-16 5:40 p.m., Carlos E. R. wrote:
>> On 2023-11-16 23:29, RabidPedagog wrote:
>>> On 2023-11-16 4:19 p.m., Carlos E. R. wrote:
>>>> On 2023-11-16 22:05, RabidPedagog wrote:
>>>>> On 2023-11-16 11:12 a.m., SBIANCO wrote:
>>>>>> Il 16/11/2023 16:46, RabidPedagog ha scritto:
>>>>>>
>>>>>>>
>>>>>>> Are you making an affirmation or an interrogation?
>>>>>>>
>>>>>>
>>>>>> I'm asking:
>>>>>> why my pc sometimes goes slow?
>>>>>
>>>>> Because you're opening an inordinate amount of tabs on four
>>>>> browsers for some ridiculous reason and expecting things to perform
>>>>> well.
>>>>
>>>> No, that on itself doesn't cause sluggishness. There are more
>>>> factors that have to concur.
>>>
>>> Well, it's rather simple. If SBIANCO isn't using a browser that
>>> blocks ads or JavaScript, every last one of those tabs is loading
>>> garbage regardless of the fact that he might not be looking at them.
>>> That's going to take a few CPU cycles no matter what he does.
>>
>> It doesn't work like that, at least in Firefox on Linux (we don't know
>> yet what he is using).
>>
>> Firefox doesn't run or load all tabs from the previous session, only
>> the one that has the focus on each window, or those you click on. The
>> rest are blank, and when clicking on them they reload from the cache.
>> Unfortunately, the only method to stop "running" tabs from running is
>> to close them or restart FF.
>>
>> There is a setting for this, I think.
>
> As far as I know, those other tabs only stop all of their processes if
> they automatically go to sleep, which is a rather new function for
> browsers. Otherwise, they're all essentially active. If Firefox puts
> them to sleep by default, and since a long time, then that's great.

What I'm telling is different.

Suppose you start FF, to reload the previous session. FF only actually
loads and runs one tab per window; the rest of the tabs are blank and
not running at all till you click on them (one by one). This is verified
behaviour in Linux.

--
Cheers,
Carlos E.R.

On 11/16/2023 11:13 AM, SBIANCO wrote:
> Il 16/11/2023 16:53, Carlos E. R. ha scritto:
>
>>
>> Enough today, but who knows in ten years? :-)
>>
>
> Surely.
> But really enough?
> Why my pc sometimes slow down?
>

SuperPI (on 6C 12T 4930K CPU)
1.5XS

Win7SP1x64 1:59.028 \
Win81x64 1:59.197 \___ I originally did these, to dispel the myth that "our best OS ever".
Win10x64 2:00.044 / When there was hardly any difference, in something that counts. Application speed.
On the other machine, even Windows XP got a respectable result.

I re-ran the tests today, and tried to make sure the Win7 result was consistent.
Then, I tested Win10 patched up to the current release. And it is 14 seconds slower.

Win7SP1x64 1:45.035 sec 8M digits of PI (boot drive SSD)
Win7SP1x64 1:44.989 sec 8M digits of PI (boot drive HDD) (I accidentally likely left Turbo on...)
Win7SP1x64 1:59.184 sec 8M digits of PI (boot drive HDD) (Turbo Off) <=== result agrees with previous test

Win10x64 2:14.922 sec 8M digits of PI (boot drive HDD) (Turbo Off) <=== NOW SLOWER, fully patched OS 19045.3693
A 14 second difference!

It is possible to benchmark things and compare them. Is what I'm saying.

The slowness could be an Intel microcode being distributed by Microsoft.
Even though my CPU is old enough (Ivy Bridge), to hardly qualify for new microcode.
It could be the Windows equivalent of "Retpoline".

The kind of slowness you're talking about, is due to the
behavior of the application. Sure, the OS could be slower,
but what the application is doing, is much more variable.

Web pages don't stay the same every day. The javascript changes.
The HTML changes. And so on.

You must be much more descriptive in your observation about "slowness",
for anyone to help you. PCs hardly ever get faster as they age.
Operations can get faster, if the code is multithreaded. When Imagemagick
started using OpenMP (parallelism), it was amazingly fast. Tested on Win10 ?
Froze Win10 tighter than a drum. Had to push the reset button. Control-alt-delete
would not work. Apparently Windows 10 at the time, did not like 12 threads
doing malloc() at the same time :-) Too funny. Microsoft fixed that.

Paul

[snip]

> Frees it? Doesn't that make ram available for use?

Available for use, rather than being used to speed up the PC.

--
38 days until the winter celebration (Monday, December 25, 2023 12:00 AM
for 1 day).

Mark Lloyd
http://notstupid.us/

"There was never such a gigantic lie told as the fable of the Garden of
Eden."

On 11/16/2023 6:27 PM, candycanearter07 wrote:
> On 11/16/23 10:01, Mark Lloyd wrote:
>> On 11/15/23 17:09, Paul wrote:
>>> On 11/14/2023 11:11 PM, Relf wrote:
>>>> David_LaRue:
>>>>> Addng RAM will help any platform.
>>>>> My 64GB for Win 7 Pro does all I need and its over 10 years old.
>>>>> I'm not a gamer though.
>>>>
>>>> What are you doing that requires 64 GB of RAM ?
>>>>
>>>
>>> RAM is never wasted. You can always put it to use.
>>
>> Its wasted if you use a program that frees it, although I haven't heard of anyone done that for years.
>>
>> [snip]
>>
>
> Frees it? Doesn't that make ram available for use?

The OS can ask applications to give up RAM they don't need.
If you put the OS under "memory pressure", it starts
casting about for unneeded RAM.

I can give a practical example with Microsoft ICE panorama software.

The program was a RAM glutton. It used RAM for a definite purpose.
But the program did not seem to know when the locality of reference
was exhausted, and the odds of the RAM being needed "soon" were
remote indeed.

if you "squeezed" the program repeatedly, it would return *30GB*
of RAM to the system. This could be doled out to... the same program.
It used the RAM slowly. Until it had grabbed the 30GB. You would
give it another kick in the nuts. And so on.

Why would you do this ? Well, the average performance of the
program would rise from 1 megabyte per second to 1.5 megabyte
per second, which just might cause your panoramic picture to be
ready a day or two sooner.

Most programs are simple enough, they know when an allocation
is no longer needed, and they free it up.

On Linux, you can do

sudo /sbin/sysctl -w vm.drop_caches=3

This method is from the old days.

https://unix.stackexchange.com/questions/17936/setting-proc-sys-vm-drop-caches-to-clear-cache

but I would not become too excited, as this really does not
fix anything. I tested, and no performance improvement was seen.
It's just another bookkeeping trick. I haven't issued one of those
in years. But there is an entry in my notes file for it.

Paul

Il 16/11/2023 22:05, RabidPedagog ha scritto:

> Because you're opening an inordinate amount of tabs on four browsers for
> some ridiculous reason and expecting things to perform well.

What hardware can manage my request?

Il 16/11/2023 22:19, Carlos E. R. ha scritto:

>
> There are more factors
> that have to concur.
>

For example?

Il 16/11/2023 23:29, RabidPedagog ha scritto:

> That's
> going to take a few CPU cycles no matter what he does.
>

RAM and CPU are ok, charge about 50%.

Il 16/11/2023 23:40, Carlos E. R. ha scritto:

> (we don't know
> yet what he is using).
>

Chrome, Edge, Firefox, Brave, Opera.

Il 16/11/2023 19:53, Carlos E. R. ha scritto:

> Windows or Linux?
>
> The second I can try to help; the former, I defer to others.
>

Windows 11

Il 17/11/2023 18:01, Paul ha scritto:

> You must be much more descriptive in your observation about "slowness",

Something like low ram.
Now I've installed TAB SUSPENDER extension in Chrome, and no problem
until now.

Windows 11.
Ryzen 5700X.
32GB RAM 3600 MHz.

On 11/16/2023 11:10 AM, SBIANCO wrote:
> Il 16/11/2023 02:52, rbowman ha scritto:
>> On Wed, 15 Nov 2023 13:08:47 -0500, RabidPedagog wrote:
>>
>>> Nobody realistically needs to open 20 tabs at a time.
>
> Yes; but I use 200 tabs, on 3 or 4 browsers.
> RAM runs good: 50% of 32GB.
> But I see some lag, some slow-down.
> What about?

There are some command line arguments you can play with.

firefox.exe --process-per-tab # launches way more processes than required
chrome.exe --process-per-site # a silly idea

about:performance
about:support
and the chrome equivalents

The purpose of starting more processes, is to reduce the number
of tabs that "are in distress", and level the CPU load across
the available cores.

Paul

On 11/17/23 14:23, Paul wrote:
> On 11/16/2023 6:27 PM, candycanearter07 wrote:
>> On 11/16/23 10:01, Mark Lloyd wrote:
>>> [snip]
>>>
>>
>> Frees it? Doesn't that make ram available for use?
>
> The OS can ask applications to give up RAM they don't need.
> If you put the OS under "memory pressure", it starts
> casting about for unneeded RAM.
>
> I can give a practical example with Microsoft ICE panorama software.
>
> The program was a RAM glutton. It used RAM for a definite purpose.
> But the program did not seem to know when the locality of reference
> was exhausted, and the odds of the RAM being needed "soon" were
> remote indeed.
>
> if you "squeezed" the program repeatedly, it would return *30GB*
> of RAM to the system. This could be doled out to... the same program.
> It used the RAM slowly. Until it had grabbed the 30GB. You would
> give it another kick in the nuts. And so on.

Oh, did it have memory leaks?
--
user <candycane> is generated from /dev/urandom

Il 18/11/2023 18:10, Paul ha scritto:

> The purpose of starting more processes, is to reduce the number
> of tabs that "are in distress", and level the CPU load across
> the available cores.
>
> Paul

Thank Yoy very much!

On 11/18/2023 11:37 AM, SBIANCO wrote:
> Il 16/11/2023 22:05, RabidPedagog ha scritto:
>
>> Because you're opening an inordinate amount of tabs on four browsers for some ridiculous reason and expecting things to perform well.
>
> What hardware can manage my request?

The software simply isn't written correctly to perform well.

Hardware cannot fix this. Hardware is defenseless against poorly written software.

If the "process-per-tab" was designed properly, it might make some tabs
work normally, while the "badly designed" web pages would be slug-slow.
And that to me, is about the best you could expect.

Paul

On 11/18/2023 11:47 AM, SBIANCO wrote:
> Il 17/11/2023 18:01, Paul ha scritto:
>
>> You must be much more descriptive in your observation about "slowness",
>
> Something like low ram.
> Now I've installed TAB SUSPENDER extension in Chrome, and no problem until now.
>
> Windows 11.
> Ryzen 5700X.
> 32GB RAM 3600 MHz.

With a tab suspender loaded in a browser, that hardware should
be plenty for the job.

There was talk of the browser itself having tab-suspender behavior,
without adding anything. I do not know if the feature is deployed yet or not.
One of the reasons the feature was designed, is to stop a single tab
from "running away" with 3GB of RAM on its own. That's why they discussed
doing one of those, not because they wanted to balance browser behavior.

It should be remembered, that the modern browsers use compositing at 60FPS.
They work on a pixmap. They update the pixmap at high frequency. The browser
does not "redraw" like it did in the old days. But compositing likely
has an impact on the graphics subsystem, and I don't really know how that
works.

https://developer.chrome.com/blog/inside-browser-part3/

Paul

On 11/18/2023 11:55 PM, candycanearter07 wrote:
> On 11/17/23 14:23, Paul wrote:
>> On 11/16/2023 6:27 PM, candycanearter07 wrote:
>>> On 11/16/23 10:01, Mark Lloyd wrote:
>>>> [snip]
>>>>
>>>
>>> Frees it? Doesn't that make ram available for use?
>>
>> The OS can ask applications to give up RAM they don't need.
>> If you put the OS under "memory pressure", it starts
>> casting about for unneeded RAM.
>>
>> I can give a practical example with Microsoft ICE panorama software.
>>
>> The program was a RAM glutton. It used RAM for a definite purpose.
>> But the program did not seem to know when the locality of reference
>> was exhausted, and the odds of the RAM being needed "soon" were
>> remote indeed.
>>
>> if you "squeezed" the program repeatedly, it would return *30GB*
>> of RAM to the system. This could be doled out to... the same program.
>> It used the RAM slowly. Until it had grabbed the 30GB. You would
>> give it another kick in the nuts. And so on.
>
> Oh, did it have memory leaks?

The program basically didn't know how to free().

There has to be a garbage collector in there somewhere,
but I couldn't determine if it was really approaching stasis
(as much malloc as free), or it was just getting slower. But
when the call by the OS to free up RAM came in, it seemed
to have no problem releasing assets then. It did not gobble up
RAM quickly. It took a while to re-consume the 30GB freed up.

It's too bad the developer of the program moved on and no longer
worked on the program. If the developer had stayed on, who knows,
we might have got an explanation of what was going on inside the
program.

The program would ask you at one point, for a "2TB hard drive
for scratch". Now, if you used a PC with 2TB of RAM, would
it have done all of that in RAM ? That, I don't know. I found
a post by one person, where it appears they were asked for a "6TB hard drive" :-)

What I was doing, is I had a "map book" where each page was
a portion of a larger map. I scanned the pages, and was then
trying to make a wall map out of the thing. I wanted to see
if you could actually scale and glue the pages together to make
a "good looking map".

What I learned, is paper has weird properties. It has a grain.
The expansion and shrinkage in the two directions is different.
Basically, if the map book, if it had been manufactured on
Tuesday, I scanned it on Wednesday, my project might have
worked. After the map book was sitting around for a year, the
paper had degraded sufficiently, I could no longer make a panorama of it.
The scales were all different. And the program didn't have a flat
transform either. It had things like Mercator projections or the like,
as a surface for tiling.

But that's probably the most unruly thing I've done on a computer.
it's like herding turtles.

Paul

Relf wrote on 15/11/2023 :
> What are you doing that requires 64 GB of RAM ?

I'd need so much RAM only in case I run 3-4-5 VM at once, but then the
CPU would be the bottleneck

--
/-\ /\/\ /\/\ /-\ /\/\ /\/\ /-\ T /-\
-=- -=- -=- -=- -=- -=- -=- -=- - -=-
............ [ al lavoro ] ...........

Paul <nospam@needed.invalid> wrote:
> On 11/18/2023 11:47 AM, SBIANCO wrote:
> > Il 17/11/2023 18:01, Paul ha scritto:
> >
> >> You must be much more descriptive in your observation about "slowness",
> >
> > Something like low ram.
> > Now I've installed TAB SUSPENDER extension in Chrome, and no problem
> > until now.
> > Windows 11.
> > Ryzen 5700X.
> > 32GB RAM 3600 MHz.
>
> With a tab suspender loaded in a browser, that hardware should
> be plenty for the job.
>
> There was talk of the browser itself having tab-suspender behavior,
> without adding anything. I do not know if the feature is deployed yet or not.

Yes, the tab-deactivation function is now part of Chrome. See the
'Memory Saver' link on this page:

'Personalize Chrome performance'
<https://support.google.com/chrome/answer/12929150>

So SBIANCO probably does not need the 'TAB SUSPENDER extension' any
longer.

FWIW, I have *not* switched on the Memory Saver feature and Chrome is
working fine with often some 20 tabs open on my 16GB Windows 11 system.

> One of the reasons the feature was designed, is to stop a single tab
> from "running away" with 3GB of RAM on its own. That's why they discussed
> doing one of those, not because they wanted to balance browser behavior.
>
> It should be remembered, that the modern browsers use compositing at 60FPS.
> They work on a pixmap. They update the pixmap at high frequency. The browser
> does not "redraw" like it did in the old days. But compositing likely
> has an impact on the graphics subsystem, and I don't really know how that
> works.
>
> https://developer.chrome.com/blog/inside-browser-part3/

Il 20/11/2023 14:27, Frank Slootweg ha scritto:

> Yes, the tab-deactivation function is now part of Chrome. See the
> 'Memory Saver' link on this page:
>
> 'Personalize Chrome performance'
> <https://support.google.com/chrome/answer/12929150>
>
> So SBIANCO probably does not need the 'TAB SUSPENDER extension' any
> longer.

BEFORE of "Tab suspender", Chrome feature was activated on my pc, and I
had problems.

NOW, with "Tab suspender", Chrome feature is activated, and I have no
problem (to this day).

On 2023-11-20 09:25, Paul wrote:
> On 11/18/2023 11:47 AM, SBIANCO wrote:
>> Il 17/11/2023 18:01, Paul ha scritto:
>>
>>> You must be much more descriptive in your observation about "slowness",
>>
>> Something like low ram.
>> Now I've installed TAB SUSPENDER extension in Chrome, and no problem until now.
>>
>> Windows 11.
>> Ryzen 5700X.
>> 32GB RAM 3600 MHz.
>
> With a tab suspender loaded in a browser, that hardware should
> be plenty for the job.
>
> There was talk of the browser itself having tab-suspender behavior,
> without adding anything. I do not know if the feature is deployed yet or not.
> One of the reasons the feature was designed, is to stop a single tab
> from "running away" with 3GB of RAM on its own. That's why they discussed
> doing one of those, not because they wanted to balance browser behavior.

I didn't know about tab suspenders. I just tried one on Firefox, but
removed it instantly: it suspends all tabs that are inactive for five
minutes. I'd like to suspend only certain tabs, manually.

What I do is open about:performance, and if there is a tab eating
resources, I consider killing it. Suspending the activity of a
particular tab would be wonderful. I don't worry about memory usage, but
cpu, which heats the laptop and makes the fan run non stop.

>
> It should be remembered, that the modern browsers use compositing at 60FPS.
> They work on a pixmap. They update the pixmap at high frequency. The browser
> does not "redraw" like it did in the old days. But compositing likely
> has an impact on the graphics subsystem, and I don't really know how that
> works.
>
> https://developer.chrome.com/blog/inside-browser-part3/
>
> Paul
>
>

--
Cheers,
Carlos E.R.

On 2023-11-20 09:43, Paul wrote:
> On 11/18/2023 11:55 PM, candycanearter07 wrote:
>> On 11/17/23 14:23, Paul wrote:

....

> What I was doing, is I had a "map book" where each page was
> a portion of a larger map. I scanned the pages, and was then
> trying to make a wall map out of the thing. I wanted to see
> if you could actually scale and glue the pages together to make
> a "good looking map".
>
> What I learned, is paper has weird properties. It has a grain.
> The expansion and shrinkage in the two directions is different.
> Basically, if the map book, if it had been manufactured on
> Tuesday, I scanned it on Wednesday, my project might have
> worked. After the map book was sitting around for a year, the
> paper had degraded sufficiently, I could no longer make a panorama of it.
> The scales were all different. And the program didn't have a flat
> transform either. It had things like Mercator projections or the like,
> as a surface for tiling.
>
> But that's probably the most unruly thing I've done on a computer.
> it's like herding turtles.

:-)

I had a script to download the tesserae of an area of google maps, as
individual photos of an area at the max resolution of the time, then
join all the photos in order to create a huge photo.

But then the photo could not be opened, gimp or whatever had out of
memory problems.

--
Cheers,
Carlos E.R.