PerfMon AutoScale Issue

[eclectic-tech]

I have a situation with using AutoScale=1 on a PerfMon measure:
My measure & meter:

Code: Select all

[mWorkingSet3]
Measure=Plugin
Plugin=Perfmon
PerfMonObject=Process
PerfMonCounter=Working Set
PerfMonInstance=[mProcess3]
PerfMonDifference=0
DynamicVariables=1

...

[TopProcess3Mem]
Meter=String
MeterStyle=sTP
MeasureName=mWorkingSet3
MeasureName2=mProcess3WS
X=([BGProcess:W]-15)r
Y=r
StringAlign=Right
Text=(WS:%2) | %1
NumofDecimals=1
AutoScale=1
DynamicVariables=1

Autoscale=1 works on 4 other identical measures and meters, but when PerfMonInstance=svchost, the scale is wrong (screenshot:3rd process, shows (TM) taskmanager value | PerfMn value (auto-scaled).
Screenshot:


Any thought on why 'Working Set' is being scaled incorrectly?

EDIT: I can set 'Scale=1000000.0' and it displays correctly, but no autoscale.

[eclectic-tech]

Since I don't feel the info from the PerfMon plugin is all that reliable, I am going in a different direction.

If anyone does find out why this is happening, I am curious, but for now, it is not an issue stopping me from finishing my lasted project...

Thanks!

[Yincognito]

Autoscale=1 works on 4 other identical measures and meters, but when PerfMonInstance=svchost, the scale is wrong (screenshot:3rd process, shows (TM) taskmanager value | PerfMn value (auto-scaled).

Wrong. Check out the FIRST svchost process from the list on the left. It has a WS of 1716, which is exactly 1.7 x 1000. So, you probably got the wrong svchost process listed in your skin (e.g. you got the 1716 WS svchost, not the 17180 one). That's my guess - I might be wrong, but it seems logical to me. Care should be taken on the svchost things, as there are a bunch of those, when these are in fact other processes using svchost to do their jobs.

EDIT: I can set 'Scale=1000000.0' and it displays correctly, but no autoscale.

If I'm right on what I've said above, this might be related to the same possible "error" in your code which lists the wrong svchost process, only this time it's got the right one. Hm... not too logical this time, but this is the only way to explain things

Since I don't feel the info from the PerfMon plugin is all that reliable, I am going in a different direction.

I'm curious why would you say that? I'm working on a similar project (being just at the last step : top processes)



and I find the perfmon info and plugin reliable enough - at least on the memory stats. What I'm having "problems" with are other stats (compared to Windows Resource Monitor) - like I/O stats, processor usage/processor user time stats for a particular process...

From what I've come to know by doing EXTENSIVE testing and checking different "utilities" against one another is that 99% of the "monitoring" software outhere (probably including Rainmeter too) rely on perfmon captures. Of course their authors would say that they've done things differently, but the facts remain: a lot of the monitoring software stats are found in almost the exact form and values in the perfmon counters (hint, hint...)

Hoping that I was helpful in identifiying the problem here, and good luck with your project.

Yin

P.S. I'm not assuming that you definitely have an error in your skin code, but, as I said above, that would offer a logical explanation for your problem.

[eclectic-tech]

Thanks for the feedback.

I did overstate my displeasure with PerfMon; you are right, that it DOES return accurate info.

The screenshot shows 3 different lists of the same processes, by PSList.exe, Windows Task Manager, and the values obtained in Rainmeter by parsing a sorted list from PSList.

I am sure it is not looking at the wrong svchost.exe; yes, there are multiple of that process, but the parsed value shown in Rainmeter is the same PID (1436), with a value of 17180k, which AutoScale says is 1.7M, instead of 17.1M

I still believe there is an issue with AutoScale, but I will leave this to better minds than mine.

Thanks again for your explanation.

[Yincognito]

I am sure it is not looking at the wrong svchost.exe; yes, there are multiple of that process, but the parsed value shown in Rainmeter is the same PID (1436), with a value of 17180k, which AutoScale says is 1.7M, instead of 17.1M

Well, this time you are right, I didn't check that, it was a quick look and noticing the 1716 WS svchost I thought this could be the culprit. My mistake.

I've never had problems with AutoScale, but then again... I prefer and use custom "scaling" in about 75%+ of the cases - I don't like the way Autoscale resizes text, I only find it useful when I need a long number to fit a certain limited/small space on the skin.

.....................

By the way, do you have an idea on how one could get top network processes (in a way similar to Windows Resource Monitor/firewall software)?
Perfmon has some very useful stats for memory, the CPU part can be done using TopProcesses/AdvancedCPU plugins or by computing the process user time counters from PerfMon (more difficult), the disk top processes can also be estimated using PerfMon counters, but the network? Having all these four groups of top processes (CPU, memory, disk and network) would be the complete package and would help identify bottlenecks in a split-second, while still looking nice and dandy using a cool Rainmeter skin...

And coupled with RivaTuner Statistics Server in games... what more could anyone want from monitoring computer usage?

[eclectic-tech]

... By the way, do you have an idea on how one could get top network processes (in a way similar to Windows Resource Monitor/firewall software)? ...

Haven't really looked too closely at network resources, but AFAIK, I do not recall seeing any plugins that currently provide that information.

Sounds like a good project for an ambitious plugin creator

[Yincognito]

Sounds like a good project for an ambitious plugin creator

Yup. Too bad I'm "only" an ambitious skin creator...for now.