Jump to content
AIDA64 Discussion Forum
Ruuwa

Whats the time interval used to calculate CPU utilization?

Recommended Posts

Greetings,

Would like to ask some questions regarding AIDA64's method of calculating the utilization value of a specific logic core in a multi core, HT enabled Intel Haswell processor. Especially about the time frame it considers. Let's assume I set an update frequency of 1 sec, so a new value will be logged/shown every second. Will it reflect the utilization based on the previous 100 ns / 100 ms / 1s? What is the time interval during which the algorithm counts system/idle time? Does it shows an averaged value of several utilization values calculated in the last 1 sec or just the last reported value?

Share this post


Link to post
Share on other sites
On ‎2017‎. ‎05‎. ‎14‎. at 7:01 PM, Ruuwa said:

Greetings,

Would like to ask some questions regarding AIDA64's method of calculating the utilization value of a specific logic core in a multi core, HT enabled Intel Haswell processor. Especially about the time frame it considers. Let's assume I set an update frequency of 1 sec, so a new value will be logged/shown every second. Will it reflect the utilization based on the previous 100 ns / 100 ms / 1s? What is the time interval during which the algorithm counts system/idle time? Does it shows an averaged value of several utilization values calculated in the last 1 sec or just the last reported value?

When you configure a 1 sec update rate, then the actual CPU utilization value will reflect the average CPU utilization over a 1 second period.

Share this post


Link to post
Share on other sites

Thank you, just one more thing. Is it the same if I set a shorter, eg 500 ms logging frequency, does it reflects the utilization of the last 500 ms even when the load and readout delay does not allow for a log entry to be produced every 500 ms? Or will it continue counting until the next readout whenever may it happen?

I guess it to be the latter from results like the one on the attached picture. The marked entries (between the magenta lines) are ~1 second away from each other, while the ones before are with about 0,5 sec. On top are the utilization values, then core frequencies, temperatures and power consumption at the bottom. Data all gathered by AIDA logging to CSV.

Is it safe to say, that because of the load by other applications, the logging gets delayed, the interval gets stretched? And it may also include some time spent on lower clock rates, because reaching TDP limits.

Thank you again for the help to interpret the collected utilization values correctly!

PS2_HCPUstats_start_enh.png

Share this post


Link to post
Share on other sites

You're right, the minimum update rate of the CPU utilization measurement in the log files is 1 second.  Do you want us to improve that rate to make the measurements more accurate on your system?

Share this post


Link to post
Share on other sites

Yes, it would be nice to have a slightly more granular utilization value, but only if the logging can keep up with the shorter intervals. I would not put it as a high priority issue, since knowing the interval to be 1 sec and independent of the preset logging frequency allows a good enough guess what happened. On a slightly other topic, can you add a sub second part to the logged time value? (HH:MM:SS,ss)? Just to see how frequently can the system produce a log entry when trying to log as frequently as possible.

 

Thank you for the answers!

Share this post


Link to post
Share on other sites
On ‎2017‎. ‎05‎. ‎22‎. at 1:03 PM, Ruuwa said:

On a slightly other topic, can you add a sub second part to the logged time value? (HH:MM:SS,ss)? Just to see how frequently can the system produce a log entry when trying to log as frequently as possible.

Thank you for the answers!

I don't think that's necessary, since you can easily estimate the update frequency by looking at the frequency of values.  So e.g. if you have 5 updates with the same HH:MM:SS value, then the actual update rate is 200 millisec.

Share this post


Link to post
Share on other sites
On 2017. 05. 22. at 1:03 PM, Ruuwa said:

Yes, it would be nice to have a slightly more granular utilization value, but only if the logging can keep up with the shorter intervals.

We'll decrease the minimum update rate to 0.1 sec (from the current 1 sec) in the next AIDA64 beta update that is rolling out later today.

  • Like 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

  • Similar Content

    • By nyrull
      Hi,
       
      First,excuse for my bad english
       
      I have a i7-5930K and wanted to know which cpu utilization is tied to which cpu cores.
       
      6 physical cores so 12 logicals.
       
      cpu1 utilization
      cpu 2 utilization
      cpu3 utilization
      Ect....
      cpu 12 utilization.
       
      so Core #1 ,Core #2 ect... is tied ti which cpu? utilization ?
       
      and when i look at logical core utilization in Sysinternals Process Explorer ,comparing value, i cant seems to find the same reading for all logical cores in Aida64
       
      Thank you !
    • By vpshockwave
      Hey guys!
      First of all, I'd like to say this is a fantastic piece of software! Good enough to convince me into buying it! I love having everything on my G15 LCD in one place!
      That said, I do have a few suggestions that I think would add some useful options.
      First, the ability to ignore virtual cores in the CPU Utilization reading. I can either monitor all 6 of my physical cores at once, or I can monitor the CPU as a whole, but when I monitor it as a whole, it factors in the virtual cores which are usually barely, if at all, being used. This has a detrimental effect of the CPU Utilization reading (it cuts it in half basically, thus making it inaccurate). The ability to ignore the virtual cores would make this more accurate in my opinion.
      Second, polling times less than a second. I realize the polling can eat up CPU, but for those of us with the CPUs that can handle it, polling times in milliseconds would be a great addition. I would love a 500 millisecond polling rate, and I know my CPU could handle it.
      Third, add the ability to drag / resize the LCD items with the mouse, rather than have to use the arrow keys or hit modify to change the length. This would make the custom LCD screens much easier to modify and create.
      Thanks for the great piece of software guys! Keep it up!
×
×
  • Create New...