Measurement of CPU usage

[PaulFXH]

Daniel Mandic escreveu:

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.

Whoa!! Daniel, take it easy, man!

I did NOT say that WinMe is better than WinXP or any other NT-based OS.
What I DID say was that, having been somewhat fearful of having to put
up with the WinMe OS for a three-month period while away from my
home-based WinXP machine, I was expecting some rough days ahead.
Actually, I was very pleasantly surprised to find that, once the
computer is cleaned up of unnecessary files, HD defragged, AVs and
Firewalls installed, non-essential start-up apps taken out, WinMe can
run very smartly for office tasks and browsing activities. In some
cases, based on my own personal experience, this WinMe machine performs
better than my WinXP box at home.
However, I don't believe anyone in their right mind would ever say that
WinMe is better than WinXP or the upcoming Vista.
Nevertheless, if this WinMe that I'm using now is as bad as it's ever
going to get, I won't have a lot to complain about.

Enjoy what's left of your weekend
Paul

 

[Frank Bohan]

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

Process Explorer will do the same, and it can be configured to replace Task
Manager, including being invoked with Ctrl-Alt-Del. As it gives more
information than WTM I prefer it.

===

Frank Bohan
¶ Accept that some days you're the pigeon, and some days you're the statue.

 

[Daniel Mandic]

going to get, I won't have a lot to complain about.

Enjoy what's left of your weekend
Paul

Just as you like.


Don't worry about. And do not take it easy, only hardest work to
knowledge, makes a fast and responsible PC.


Wish you good luck with your testings... As I said, I would look only
to orig. Tools, or some sysinternals stuff.




Best Regards,

Daniel Mandic

 

[Al Klein]

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%.

Nope - the CPU is ALWAYS executing *some* instruction (unless it
executed a HALT instruction, in which case it's stopped), so usage is
always 100%. Not 99,9%, not 100.1% - always exactly 100%. Which is
why measuring total CPU usage doesn't tell you anything.

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).

As I said, that's impossible - the CPU is ALWAYS executing an
instruction.

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.

That would be nice, but it won't work.

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.

No, since the CPU, if you define "busy" as "executing instructions",
is always busy. At least until you turn the computer off.

If the program is giving you a measure of
"cpu-in-use-but-not-executing-System Idle Process", it may be
accurate. But if it's claiming that the CPU is only being used 60-80%
of the time, it's lying. If the CPU is doing "nothing", it's
executing a wait loop (called System Idle Process), waiting for
something to happen.

 

[Jast]

Jast escreveu:


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

I downloaded and installed it on my XP box and it is quite close to what
process explorer is indicating, usually within 10% of process explorer,
and usually higher. I then set the update speed to 5secs on both of them
and this made both cpu measurements within 3% or less of each other. So,
on my box they both agree with each other. I also tried it on various
cpu usages from idle to around 60% up to 95% and once again they were
both consistent.

 

[PaulFXH]

Jast escreveu:

I downloaded and installed it on my XP box and it is quite close to what
process explorer is indicating, usually within 10% of process explorer,
and usually higher. I then set the update speed to 5secs on both of them
and this made both cpu measurements within 3% or less of each other. So,
on my box they both agree with each other. I also tried it on various
cpu usages from idle to around 60% up to 95% and once again they were
both consistent.

Hi Jast
Thanks a lot for trying out that little test.
However, you have really pulled the carpet with your results. Maybe I
am crazy, after all!
Seriously, I'm absolutely flabbergasted as I have already seen the
behaviour I describe on not just this machine but ANOTHER running on
WinXP SP2.
On my present box, however, Process Explorer indicated about 3-5% CPU
Usage under normal unstressed situations. If I add 10% to this (as you
suggest) I come up with 13-15% CPU Usage indicated by FreeMeter which
is actually a very big difference (athough still a lot less than the
difference I got).
Also if CPU usage is very heavy (close to 100%) then both meters
necessarily must be both very close to 100%.
It would help me understand a little better the significance of your
tests if you were able to supply the ACTUAL numbers you got; e.g. Test
1: ProcExp=X, FreeM=Y and so on for the numbers of simultaneous
comparisons you made.

Is there anybody else out there who might want to try out this simple
test already tried by Jast (compare CPU Usage indications from Process
Explorer and FreeMeter with both utilities running at the same time)
and post the results?
It would be very interesting to get a range of results.

Thanks again, Jast
Paul

 

[Jast]

I don't want to spend hours plotting numbers. If you can give exact
instructions that take a few minutes I could possibly post the results,
but otherwise you will have to take my word for it I guess.

When I mouse over both icons of both programs in the systray, the cpu
usage is fairly similar at a range of different values when set to a 5
second sample period.

 

[Daniel Mandic]

Hi Paul!




Freemeter only have an install... It's not an archive, so I can't start
it w/o installing.

But I can say you so much. My CPU Time falls to 0.... (nothing
running), and even when (very slightly) I move the TaskManager window,
it stays at 0. What now? Is my CPU damaged? I use and it doesn't show!?





Take it easy... There is always two ways (and more, of couerse)... with
Computer things.

You haven't seen the posting from an other subscriber? I think he
tested them on XP and it was nearly the same.






Best Regards,

Daniel Mandic

 

[PaulFXH]

Jast escreveu:

I don't want to spend hours plotting numbers. If you can give exact
instructions that take a few minutes I could possibly post the results,
but otherwise you will have to take my word for it I guess.

Hi Jast
Thanks for your reply.
No, I'm not asking you to do any plotting of numbers nor, indeed, any
extra testing. How many minutes it takes depends on how good is your
memory.
I'm just asking you to provide me with the actual CPU usage numbers you
got when you did your test yesterday. So, instead of telling me that at
first there was a difference of 10% between the Proc Exp result and
that of FreeMeter, can you say if the real numbers were closer to
1%/11% or to 85%/95%. And then just add whatever other pairs of numbers
you can remember from your test.
Also, I don't think we need pinpoint precision here.

When I mouse over both icons of both programs in the systray, the cpu
usage is fairly similar at a range of different values when set to a 5
second sample period.

As a matter of interest, just before I wrote this I checked my machine
again and, with very little running other than the two utilities, Proc
Exp indicated about 2.7% CPU usage (exclusive of System Idle Processes)
while FreeMeter indicated 81% (average).
Then I did an AV scan with AntiVir and the CPU usages rose immediately
to about 50% for Proc Exp and a straight, unvarying 100% for FreeMeter.
When AntiVir was shutdown, both utilities reverted immediately to the
previous values.

Paul

 

[PaulFXH]

Daniel Mandic escreveu:

Hi Paul!




Freemeter only have an install... It's not an archive, so I can't start
it w/o installing.

Hi Daniel
Thanks for your comments.
Do you have fears about installing Freemeter on your computer? It's
very small (< 1MB), very easy to uninstall and has been recommended
many times in this NG. Also I have never seen a complaint about it.
Anyway, only you can decide if you want to install it or not; but if
you do decide, you can get it from here:
http://www.tiler.com/freemeter/download/license.php

But I can say you so much. My CPU Time falls to 0.... (nothing
running), and even when (very slightly) I move the TaskManager window,
it stays at 0. What now? Is my CPU damaged? I use and it doesn't show!?


Does your indication of CPU usage stay at zero when you run, for example, an AV scan or an AVI (movie)? If so, it would seem that Task Manager is not picking up the relevant information for CPU indication possibly due to corruption of a system file.

If your CPU itself were damaged I don't think you would be seeing
anything at all, not just the CPU indication on your TM.

Take it easy... There is always two ways (and more, of couerse)... with
Computer things.

You haven't seen the posting from an other subscriber? I think he
tested them on XP and it was nearly the same.

Yes, I did see Jast's results and have replied to him. I am quite
startled by this revelation. However, as I have already posted, I did
actually see the same type of difference in CPU usage indication on my
home computer which runs on WinXP. However, I'm a long way from home
right now so can't recheck this.

Cheers
Paul

 

[PaulFXH]

Daniel Mandic escreveu:

Hi Paul!




Freemeter only have an install... It's not an archive, so I can't start
it w/o installing.

BTW, Daniel, if you still don't want to install FreeMeter on your
computer, there is a possibility that you already have a utility that
will give you an estimate of CPU usage. Incidentally, ALL of these
utilities give the very high indications of CPU usage that I have seen
with FreeMeter.
These are:
1. Aida32 v. 3.93
2. PCWizard v. 1.68
3. Everest (last available FreeWare version)

All of these are free and provide a massive amount of information
(largely hardware, but some software too) about your computer. All
include an estimate of CPU usage.

Paul

 

[Daniel Mandic]

BTW, Daniel, if you still don't want to install FreeMeter on your
computer, there is a possibility that you already have a utility that
will give you an estimate of CPU usage. Incidentally, ALL of these
utilities give the very high indications of CPU usage that I have seen
with FreeMeter.
These are:
1. Aida32 v. 3.93
2. PCWizard v. 1.68
3. Everest (last available FreeWare version)

Yes, no problem Paul.


Please know, there is so much what could cause your phenomenon....


Maybe next time... This Tiler installs me too much and needs a restart
too, so far I saw.

Something out of the file I will test!
The most apps can be started from a folder.




Best Regards,

Daniel Mandic

 

[Jast]

Ok here are the numbers, idle
1 2
1 2
4 2
3 2
2 3
4 2
Then I started playing a .mov file
36 29
26 26
26 27
28 26
Then I also started playing an avi
70 75
77 77
76 74
63 58

I had them both on 5 second sampling. I have an XP Athlon 1800+ CPU.

 

[PaulFXH]

Jast escreveu:

Ok here are the numbers, idle
1 2
1 2
4 2
3 2
2 3
4 2
Then I started playing a .mov file
36 29
26 26
26 27
28 26
Then I also started playing an avi
70 75
77 77
76 74
63 58

Hi Jast
Thanks a lot for carrying out the tests.
Well, it sure looks like a pretty conclusive proof that FreeMeter and
Process Explorer do indeed give the same CPU usage numbers.

So, why are these numbers so astoundingly different on my computer? Is
it because I have WinMe rather than WinXP? But certainly FreeMeter
claims to be compatible with WinMe.

Is there some hardware/software deficiency on my machine that prevents
accurate CPU usage measurement? But at least two utilities (Process
Explorer and WinTop) do give what I would consider "trustable" results.

I'm going to have to scratch my head for a week or so to see where to
go next.

Many thanks for your help.
Paul

 

[Daniel Mandic]

Is there some hardware/software deficiency on my machine that prevents
accurate CPU usage measurement? But at least two utilities (Process
Explorer and WinTop) do give what I would consider "trustable"
results.

Yes, Win9x/ME.






Please trim your postings... It's a horror to find the related text.

E.g.: Look at this posting here! Cut unrelated and obsolete!!





Best Regards,

Daniel Mandic

 

[PaulFXH]

Daniel Mandic escreveu:

Yes, Win9x/ME.






Please trim your postings... It's a horror to find the related text.

E.g.: Look at this posting here! Cut unrelated and obsolete!!

Well, actually Daniel, I use Google Groups as my NG viewer and ONLY the
"new" text and the question to which it refers can be seen
What you refer to as "unrelated and obsolete" text remains hidden under
links. So, you can see it if you want or choose to leave it hidden.
This saves having to cut away stuff.
How are viewing your news?
Paul

 

[Daniel Mandic]

How are viewing your news?
Paul

YYes.


Try out some freeware News-clients.

Xnews, XanaNews etc... Freeagent and so.



Best Regards,

Daniel Mandic

 

[Sietse Fliege]

Yes, Win9x/ME.






Please trim your postings... It's a horror to find the related text.

E.g.: Look at this posting here! Cut unrelated and obsolete!!





Best Regards,

Daniel Mandic

Please, All those blank lines... It's a horror to find text in between.

E.g.: Look at this posting here! Cut the obsolete!!

 

[Josef W. Segur]

Jast escreveu:


Hi Jast
Thanks a lot for carrying out the tests.
Well, it sure looks like a pretty conclusive proof that FreeMeter and
Process Explorer do indeed give the same CPU usage numbers.

So, why are these numbers so astoundingly different on my computer? Is
it because I have WinMe rather than WinXP? But certainly FreeMeter
claims to be compatible with WinMe.

I have read that Win NT/2k/XP have a HLT instruction in their idle process
so the CPU is actually stopped when not needed. On Win9X/ME systems the
idle process is an active loop. That major difference is probably what is
causing the discrepancy.

There are "CPU Cooler" programs which can be used on Win9X/ME systems to
add the HLT. <URL: http://www.benchtest.com/cooler.html > might provide
some ideas.

 

[PaulFXH]

Josef W. Segur escreveu:

I have read that Win NT/2k/XP have a HLT instruction in their idle process
so the CPU is actually stopped when not needed. On Win9X/ME systems the
idle process is an active loop. That major difference is probably what is
causing the discrepancy.

There are "CPU Cooler" programs which can be used on Win9X/ME systems to
add the HLT. <URL: http://www.benchtest.com/cooler.html > might provide
some ideas.

Hi Joe
Thanks for your comments.
That's an interesting observation about a difference in CPU activity
between Win9x/Me and NT OSs. However, I cannot as yet see how this
would lead to two different numeric assessments of CPU activity on
machines with Win9x/Me.
Since I initiated this thread, I have learnt that FreeMeter and Process
Explorer give essentially identical CPU Usage readings on machines
running WinXP (based on tests with two different computers).
Nevertheless, my own computer on WinMe continues to show a very marked
difference (FM number is normally 10-20 times that of PE) between the
indicated CPU Usage for these two utilities.
I still don't know if this is due to some peculiarity of my computer or
whether it is a characteristic of machines running on Win9x/Me.
Is there anybody out there running any of these non-NT OSs who would be
prepared to try a small, free, risk-free test on CPU Usage?

TIA
Paul