Measurement of CPU usage

[PaulFXH]

Hi
Not looking for FreeWare--just trying to hear some opinions on
something that been bothering me for some time.
I use WinMe as OS and have downloaded various small utilities (all
FREE) to view CPU Usage on my computer.
These included Process Explorer, ActiveCPU, FreeMeter and WinTop.
Trouble is, they don't all show the same result. Indeed, the CPU Usage
results seem to fall into one of two distinct categories; very high or
very low.
As an example, while writing this I have all four of the above
utilities running showing the CPU usage (note that I have only one
single-core processor on this box). So, while Process Explorer and
WinTop show about 3.5% usage, FreeMeter and ActiveCPU, on the other
hand, both show about 64%.
I have seen exactly the same thing on at least one other computer
(running on WinXP).
I have posted to a number of groups trying to get somebody to explain
why there is this large difference and what significance should be
attached to it. I have even e-mailed the author of one of the
utilities.
However, up to now, nobody seems able, or willing, to provide an
explanation.
Is there anybody out there who can point me in the right direction on
this one?
TIA
Paul

 

[Daniel Mandic]

Is there anybody out there who can point me in the right direction on
this one?
TIA
Paul

NT. (or try some frree Unix, although I don't know the Linux CPU-Task
bheaviour! In my case I can see everything with Ctrl-Alt-Del
'Taskmanager')


Best Regards,

Daniel Mandic

 

[PaulFXH]

Daniel
Thanks for your reply.
However, I'm NOT looking for a means to see what my CPU usage is--I
already have about 6 utilities showing me this.
What I want is an explanation as to why these utilities provide such
startlingly different results.
Which one is right?
Or, are they both right with different interpretations?

Paul

 

[Daniel Mandic]

Or, are they both right with different interpretations?

Paul

I Don't Know.

I think I saw a Taskmanager for 98SE, in one of these (unofficial)
Service-Packs. (discussed some days/weeks ago!)

Maybe there (that) is a better CPU-meter!?




Best Regards,

Daniel Mandic

P.S.: I think I tried this one:
http://www.bpa23.host.sk/wupg.html

Here is ME: http://www.mdgx.com/web.htm

 

[PaulFXH]

Daniel Mandic escreveu:
Daniel
Thanks once again for your comments and the links.

I Don't Know.

I think I saw a Taskmanager for 98SE, in one of these (unofficial)
Service-Packs. (discussed some days/weeks ago!)

However, I really don't want, or need, to download anything. Indeed,
the WinTop which I use for CPU Usage indication (among others) I got
from WinMe Kernel Toys. This is a genuine Windows product specifically
for WinMe which does a lot, but not all, of the things that Taskmanager
does in the NT OSs.
So, a Taskmanager substitute for WinMe is unlikely to help me at all.

Maybe there (that) is a better CPU-meter!?

But to know if something is BETTER, I need first of all to know what is
WRONG with the utilities that I already have.
Indeed, this is the sole purpose of my post.
Essentially, is the (generally) low CPU Usage figure indicated by
Process Explorer/WinTop more realistic/believable than the (generally)
much higher CPU Usage number indicated by ACPU/FreeMeter?

TIA
Paul

 

[hummingbird]

Hi
Not looking for FreeWare--just trying to hear some opinions on
something that been bothering me for some time.
I use WinMe as OS and have downloaded various small utilities (all
FREE) to view CPU Usage on my computer.
These included Process Explorer, ActiveCPU, FreeMeter and WinTop.
Trouble is, they don't all show the same result. Indeed, the CPU Usage
results seem to fall into one of two distinct categories; very high or
very low.
As an example, while writing this I have all four of the above
utilities running showing the CPU usage (note that I have only one
single-core processor on this box). So, while Process Explorer and
WinTop show about 3.5% usage, FreeMeter and ActiveCPU, on the other
hand, both show about 64%.
I have seen exactly the same thing on at least one other computer
(running on WinXP).
I have posted to a number of groups trying to get somebody to explain
why there is this large difference and what significance should be
attached to it. I have even e-mailed the author of one of the
utilities.
However, up to now, nobody seems able, or willing, to provide an
explanation.
Is there anybody out there who can point me in the right direction on
this one?

I can only suggest you e-mail the authors of each and explain it to
them collectively (ie to all of them in one e-mail). They might offer
an explanation.

 

[PaulFXH]

hummingbird escreveu:

I can only suggest you e-mail the authors of each and explain it to
them collectively (ie to all of them in one e-mail). They might offer
an explanation.

Good suggestion, hummingbird, for which thanks.
Note, however, I have already mailed one author without response.
Too, I really cannot believe that nobody in the NG arena has noticed
this phenomenon and come up with a plausible explanation.
I'll wait and see if anybody can suggest something here before
resorting to your recommendation.

Paul

 

[Jast]

Hi
Not looking for FreeWare--just trying to hear some opinions on
something that been bothering me for some time.
I use WinMe as OS and have downloaded various small utilities (all
FREE) to view CPU Usage on my computer.
These included Process Explorer, ActiveCPU, FreeMeter and WinTop.
Trouble is, they don't all show the same result. Indeed, the CPU Usage
results seem to fall into one of two distinct categories; very high or
very low.
As an example, while writing this I have all four of the above
utilities running showing the CPU usage (note that I have only one
single-core processor on this box). So, while Process Explorer and
WinTop show about 3.5% usage, FreeMeter and ActiveCPU, on the other
hand, both show about 64%.
I have seen exactly the same thing on at least one other computer
(running on WinXP).
I have posted to a number of groups trying to get somebody to explain
why there is this large difference and what significance should be
attached to it. I have even e-mailed the author of one of the
utilities.
However, up to now, nobody seems able, or willing, to provide an
explanation.
Is there anybody out there who can point me in the right direction on
this one?
TIA
Paul

I guess it depends on how you are testing them? You should have a
standard load set up, say a big avi file or something that maxes out
your cpu at around 50 to 90% then monitor this using only one of your
CPU monitors. Then do the same thing with each one going. Running all of
them at once will produce different results because they are sampling at
different times and thus different results. Also, try increasing the
sample period or update speed to something bigger, say at least 2 to 5
seconds as this may average over a longer time and thus give better,
smoother more accurate results.

 

[Al Klein]

However, I'm NOT looking for a means to see what my CPU usage is--I
already have about 6 utilities showing me this.
What I want is an explanation as to why these utilities provide such
startlingly different results.
Which one is right?
Or, are they both right with different interpretations?

CPU usage is always 100% - the CPU doesn't (usually) stop. The
question is which program is using what percentage of the CPU's time.

If you're looking at a program that claims to show actual CPU usage,
unless it's claiming 100% it's wrong.

If you're looking at a program that claims to show the percentage of
CPU time used by each program, be aware that usage is dynamic - it
varies microsecond by microsecond. While the system idle process (the
loop in the OS that waits for something to happen) usually gets most
of the CPU time, it gives its time up whenever another program needs a
slice. Unless both programs are reporting the same thing at the same
time (and that's impossible, due to the nature of digital computers),
they're going to give you different results. A single program will
give you different results moment by moment.

 

[PaulFXH]

Jast escreveu:
Hi Jast
Thanks a lot for your reply.
See my comments below.

I guess it depends on how you are testing them? You should have a
standard load set up, say a big avi file or something that maxes out
your cpu at around 50 to 90% then monitor this using only one of your
CPU monitors.

The problem here, Jast, is that I already have an indication of CPU
Usage in the range suggested by you. Thus, FreeMeter and ACPU were
reading 64% (more or less) when I did the test. Interestingly, this was
WITHOUT anything very big running, certainly no AVIs.

Then do the same thing with each one going. Running all of
them at once will produce different results because they are sampling at
different times and thus different results.

OK, but the difference that I mentioned in my first post was 3.5% Vs.
64%. That's very nearly a factor of 20. I don't think you need to be
mathematical genius to see that this is not due to different sampling
times.

Also, try increasing the
sample period or update speed to something bigger, say at least 2 to 5
seconds as this may average over a longer time and thus give better,
smoother more accurate results.

Actually, the sampling time, in those where it can be changed, is right
now 2 seconds. Even so, were the difference between the various results
4% Vs. 4.5%, I think we might be able to attribute it to a sampling
problem.
But 3.5% Vs. 64%, I thing we can safely rule out sampling as an
explanation for this astronomical difference.

Thanks again
Paul

 

[PaulFXH]

Al Klein escreveu:
Hi Al
Thanks for your reply.
You make some very interesting points on which I have commented below.

CPU usage is always 100% - the CPU doesn't (usually) stop. The
question is which program is using what percentage of the CPU's time.

Yes, this is my understanding too. Indeed, the two utilities that gave
the lower results in my test (Process Explorer and WinTop both gave
around 3.5%) do assume 100% total CPU usage with what's not being used
by running processes being attributed to System Idle Processes.

If you're looking at a program that claims to show actual CPU usage,
unless it's claiming 100% it's wrong.

This would force you to conclude that the CPU Usage results from
FreeMeter and ACPU are both just WRONG.
However, it's not just these two utilities that show these alarmingly
"high" CPU usage numbers (after discounting System Idle usage). Others
that give results in the same category (on my computer anyway) are
Everest, PCWizard and SysMon1.22.
I find it hard to accept that all of the utilities are just churning
out nonsense results, particularly, when they all give nonsense results
that are practically the same (in tests where I have run many of these
utilitie simultaneously).
It seems to me that these two categories of CPU Usage results derive
from different methods of measurements, or something................
However, a more precise explanation seems to be incredibly elusive.

If you're looking at a program that claims to show the percentage of
CPU time used by each program, be aware that usage is dynamic - it
varies microsecond by microsecond. While the system idle process (the
loop in the OS that waits for something to happen) usually gets most
of the CPU time, it gives its time up whenever another program needs a
slice. Unless both programs are reporting the same thing at the same
time (and that's impossible, due to the nature of digital computers),
they're going to give you different results. A single program will
give you different results moment by moment.

I agree but the differences I'm seeing in the CPU Usage numbers
(discounting the System Idle time) is absolutely enormous---3.5% vs 64%
in the test I carried out earlier today.
I don't think anybody can claim this is due to differences in
measurement times.

This was an interesting post but it seems there's still some way to go
before this strange phenomenom gets explained.
Thanks
Paul

 

[Daniel Mandic]

Too, I really cannot believe that nobody in the NG arena has noticed
this phenomenon and come up with a plausible explanation.

Phenomenon? I thought Windows 9x is already a Phenomenon.



Best Regards,

Daniel Mandic

 

[Al Klein]

I agree but the differences I'm seeing in the CPU Usage numbers
(discounting the System Idle time) is absolutely enormous---3.5% vs 64%
in the test I carried out earlier today.
I don't think anybody can claim this is due to differences in
measurement times.

Run Task Manager - sort by CPU and you'll see what's using what
percentage of CPU time.

 

[Jast]

Jast escreveu:

Ok

Personally I would be more inclined to trust process explorer. Does the
task manager in win98 show a graph like it does in winxp? If so just
compare this graph against all the other ones.

 

[Jast]

I have been using process explorer for years and it is quite adequate
for my purposes in that when ever it says my cpu is under load, my
computer does slow down. When it says my computer is idle then things
are quick.

One question on process explorer I have is when it displays red, is that
related to disk usage?

 

[PaulFXH]

Daniel Mandic escreveu:

Phenomenon? I thought Windows 9x is already a Phenomenon.

Not sure in what sense you say that my WinMe is a "phenomenon".
However, as a long time WinXP user, I have just had the opportunity
over the last 7 weeks to try out this computer running on WinMe.
I didn't expect very much particularly with the reputation this OS has
gained for itself.
But, I'm actually quite impressed and, now that I've been able to tune
it up quite a bit, this computer (256MB RAM) actually does some things
faster than my larger/faster/more RAM box running on WinXP SP2.
OK, WinMe won't accept a lot of more modern software but for somebody
who just wants to do e-mailing, browsing and some office tasks I might
even say that WinMe is better (read faster) than WinXP.
Paul

 

[PaulFXH]

Jast escreveu:

Ok

Personally I would be more inclined to trust process explorer. Does the
task manager in win98 show a graph like it does in winxp? If so just
compare this graph against all the other ones.

WinTop in WinMe, which is the equivalent of Task Manager in WinXP, does
not show a graph. On the other hand, both FreeMeter and ACPU do show
graphs.
Not sure, however, how this helps as, for example, ACPU shows a line
graph which is exactly equivalent of the line graph in Process Explorer
except that it's graphing its higher numbers.
Paul

 

[Daniel Mandic]

Daniel Mandic escreveu:


Not sure in what sense you say that my WinMe is a "phenomenon".
However, as a long time WinXP user, I have just had the opportunity
over the last 7 weeks to try out this computer running on WinMe.
I didn't expect very much particularly with the reputation this OS has
gained for itself.
But, I'm actually quite impressed and, now that I've been able to tune
it up quite a bit, this computer (256MB RAM) actually does some things
faster than my larger/faster/more RAM box running on WinXP SP2.
OK, WinMe won't accept a lot of more modern software but for somebody
who just wants to do e-mailing, browsing and some office tasks I might
even say that WinMe is better (read faster) than WinXP.
Paul

Hi Paul!



You might not have the sense, or you do not want to see the difference.
Well, XP is heavy OK, but even then.... it's a NT Kernel.

256MB is a first argument (as I don't know the machine you are talking
about, CPU, Chipset?). I would max. install NT4, with licennse, or
Windows 95,98 or ME - with 256MB.

Try out NT4... that's a blaze, compared to the 18bitters (9x/ME).
Dazzling is gone, fast true 32bit execution (due to better-kernel away
from 8 and 16bit).

IMO, Win9x/ME is more for compatibility-types... A today Windows-Spezi
(as me ;-)) goes with something NT, or Unix (non-windows
internet-specialists).


Only with NT5 and above you get the real out of your IBM-PC compatible
machine. e.g. AGP.

Win9x can start faster, open (some!) apps faster etc.. but it does not
have a memory-protection and it works in a trivial (crashs??) protected
mode. Although, the second time I start the same app on a XP (much much
much better caching strategies - crashfree), it goes faster than any
9x/ME, indeed.

Multitasking is far better. I would even say, that's (NT) the first
real multitasking-OS by Microsoft.




So, your CPU-meter results might be blurred by the OS. Try the same on
NT, and compare 'em to Ctrl-Alt-Del.

Look to Sysinternals, they have nice utils and tools for Windows.

www.sysinternals.com





Kind Regards,

Daniel Mandic

 

[PaulFXH]

Jast escreveu:

I have been using process explorer for years and it is quite adequate
for my purposes in that when ever it says my cpu is under load, my
computer does slow down. When it says my computer is idle then things
are quick.

Yes I trust Process Explorer too. Its just that I was trying to
understand why many other utilities show CPU Usage figures that are
dramatically higher.
Do you want to see exactly what I mean?
You can download and install FreeMeter (for Free) from this site:

http://www.tiler.com/freemeter/download/license.php

This is an entirely trustworthy site that has been recommended many
times in this group (and with which I have absolutely no connection
whatsoever).
The file size is 888KB which will take seconds to download if you have
a DSL connection.
Yoy could then see immediately how the CPU Usage shown by FreeMeter is
so very different from that of Process Explorer.
FreeMeter can then be very easily uninstalled through Control
Panel>Add/Remove Programs if you don't want to keep it.

One question on process explorer I have is when it displays red, is that
related to disk usage?

I had thought that the red highlighting of processes in the Process
Explorer list showed processes that were just about to shut themselves
down. In my experience, if you keep watching them (the ones highlighted
in red) they'll disappear.
Is your experience different?

Paul

 

[PaulFXH]

Al Klein escreveu:

Run Task Manager - sort by CPU and you'll see what's using what
percentage of CPU time.

Thanks for your suggestion, Al.
Problem here is the utilities that show the very high CPU Usage do NOT
(not one of them) provide any indication of what processes are using
how much of the CPU. Therefore, what you suggest is not useful to try
to figure out how realistic are the 60-80% CPU usage figures generated
by utilities such as FreeMeter, ACPU, Everest, PCWizard and SysMon1.22

However, looking at the FreeMeter homepage, it describes what it calls
"CPU Usage" as the "percentage of time that the CPU is busy".
You commented earlier that CPU usage should always be close to 100%.
If we assume that in fact it's not normally that close in reality and
that the 60-80% numbers churned out by FreeMeter, (among others) are
truly what they claim, i.e. the percentage of time that the CPU is busy
(including doing System Idle Processes).
Then, on the other hand, if we suppose that Process Explorer (and
WinTop and Task Manager) distributes the % of the CPU use to the
various processes currently running but ONLY FOR THE TIMES THE CPU IS
BUSY, then we would have an explanation.
For this reason, the total of all the CPU time being used by the
various running processes in Process Explorer is always close to
100%---because it only makes the calculation WHEN THE CPU IS BUSY.
Indeed, it seems reasonable that it can only make this calculation when
the CPU is busy.
I would welcome any views you might have on this rather convoluted
attempt at elucidating this perhaps not so complex problem.
Paul