Intel Power Gadget Help Please

[getrealbro]

Despite copious Googling, I’m having trouble understanding some of the Intel Power Gadget v3.6.1 curves.

The Power, Temperature and Utilization graphs are pretty easy to figure out. But the Frequency graph — not so much. I guessing that these curves are…

GFX curve = freq of the Intel UHD Graphics 630
Straight Line = Max freq of the cores without TurboBoost
MAX curve = top of the blue shaded area
MIN curve = bottom of shaded area
AVG curve = blue curve inside the shaded area

BUT… what exactly do the MAX, MIN and AVG represent? IOW do the MAX, MIN and AVG curves represent the frequency of the fastest, slowest and median of the 6 cores at any given time? And if so, why do these curves quickly converge above the nominal core frequency even when the Utilization curve barely budges?

Thanks in Advance — GetRealBro

 

[Spectrum]

Despite copious Googling, I’m having trouble understanding some of the Intel Power Gadget v3.6.1 curves.

The Power, Temperature and Utilization graphs are pretty easy to figure out. But the Frequency graph — not so much. I guessing that these curves are…

GFX curve = freq of the Intel UHD Graphics 630
Straight Line = Max freq of the cores without TurboBoost
MAX curve = top of the blue shaded area
MIN curve = bottom of shaded area
AVG curve = blue curve inside the shaded area

BUT… what exactly do the MAX, MIN and AVG represent? IOW do the MAX, MIN and AVG curves represent the frequency of the fastest, slowest and median of the 6 cores at any given time? And if so, why do these curves quickly converge above the nominal core frequency even when the Utilization curve barely budges?

Thanks in Advance — GetRealBro

Green is the GPU.
Straight grey line is the 'guaranteed', non turbo base GPU frequency at processor TDP (e.g. 28W, 45W, 65W, 95W, etc). Your chip is a 65W CPU based on the grey line in the power chart.
REQ is the requested frequency - normally the maximum value possible depending on number of cores being loaded - and proportioanl to the CPU specs listed on ARK.
I have assumed that the MAX, MIN and AVG reperesent the maximum, minimum and average fequency observed in each sampling window (e.g. 1 second). But, yes...I'm not sure how the multiple cores (each potentially at a different frequency) come into play here.

It looks like you have either an i7 iMac or i7 mac mini, is that correct?

 

[getrealbro]

Green is the GPU.
Straight grey line is the 'guaranteed', non turbo base GPU frequency at processor TDP (e.g. 28W, 45W, 65W, 95W, etc). Your chip is a 65W CPU based on the grey line in the power chart.
REQ is the requested frequency - normally the maximum value possible depending on number of cores being loaded - and proportioanl to the CPU specs listed on ARK.
I have assumed that the MAX, MIN and AVG reperesent the maximum, minimum and average fequency observed in each sampling window (e.g. 1 second). But, yes...I'm not sure how the multiple cores (each potentially at a different frequency) come into play here.

It looks like you have either an i7 iMac or i7 mac mini, is that correct?

Thanks for the info. "the MAX, MIN and AVG represent the maximum, minimum and average frequency observed in each sampling window" makes sense.

Yes this a "Black Friday" 2018 Mac Mini i7 w/32GB Ram (self installed).I was planning on getting an i5 but made an impulse upgrade based on the larger discount. "But look honey how much more money I saved."

I also added the Satechi aluminum stand (Cyber Monday 25% off) and noticed the Mini seemed to be hotter than I remembered before I added the stand. Hence the search for some way to monitor/record the temps, which lead to Intel Power Gadget and Macs Fan Control. Of course then I began wondering why the freq curves converge to above 3.2Ghz on almost any activity e.g. scrolling an Excel preadsheet. Even with your interpretation of the freq curves, it looks like all 6 cores would have to be in turbo boost at some point during that sample period.

GetRealBro

 

[Spectrum]

Of course then I began wondering why the freq curves converge to above 3.2Ghz on almost any activity e.g. scrolling an Excel preadsheet. Even with your interpretation of the freq curves, it looks like all 6 cores would have to be in turbo boost at some point during that sample period.

GetRealBro

Sounds like a good buy. If you want to monitor what each core is up to, use activity monitor and the CPU history chart. few processes work multiple cores at once. The cores mostly help when you are doing multiple different things at once.

so...I agree with you. These can't represent the average frequency across the 6 cores. For example, early on in the trace, you have a spike average of over 4 GHz. But the CPU utilisation is only about 15%, which is basically 2/12 cores If you include hyper threading.
[automerge]1575932127[/automerge]
EDIT. Also a full 6 core turbo will briefly use around 85Watts of power.

 

[cosmolv]

Stay away from this crappy software! I just spended about a hour to get my mac from brick to normal working condition. First it cannot boot after installing of this. Then i managed to boot into osx, but it runs slow. How such big company managed to made a such crap software?

 

[frou]

I have assumed that the MAX, MIN and AVG reperesent the maximum, minimum and average fequency observed in each sampling window (e.g. 1 second).

This can be confirmed as the right answer by going into the Power Gadget's settings and slamming both the update intervals to be as often as possible. Then from looking at the resultant graph we can see that the clockspeed is a lot more dynamic than the graph based on the default settings suggested.


(Also, the lag that exists between the OS requesting some frequency and the processor delivering it becomes more apparent).

But, yes...I'm not sure how the multiple cores (each potentially at a different frequency) come into play here.

It seems we can simplify our mental model of e.g. the Coffeelake in the Mini:

https://apple.stackexchange.com/a/364583

I'm the owner of Intel Power Gadget on macOS. [...] On any Intel processor prior to Icelake, all cores run at the same frequency. [...]

 

[sergioarista]

Thanks for the info. "the MAX, MIN and AVG represent the maximum, minimum and average frequency observed in each sampling window" makes sense.

Yes this a "Black Friday" 2018 Mac Mini i7 w/32GB Ram (self installed).I was planning on getting an i5 but made an impulse upgrade based on the larger discount. "But look honey how much more money I saved."

I also added the Satechi aluminum stand (Cyber Monday 25% off) and noticed the Mini seemed to be hotter than I remembered before I added the stand. Hence the search for some way to monitor/record the temps, which lead to Intel Power Gadget and Macs Fan Control. Of course then I began wondering why the freq curves converge to above 3.2Ghz on almost any activity e.g. scrolling an Excel preadsheet. Even with your interpretation of the freq curves, it looks like all 6 cores would have to be in turbo boost at some point during that sample period.

GetRealBro

Yes the satechi stand interferes a little with airflow; my fix was to put some tabs in the corners to raise my mm a few milimiters, it worked pretty wel...

 

[getrealbro]

Yes the satechi stand interferes a little with airflow; my fix was to put some tabs in the corners to raise my mm a few milimiters, it worked pretty wel...

I put four small rubber washers between the Satechi stand and the MM, too. It helps but the i7 still runs hotter to the touch than I would like.

I’ve recently been transcoding some real time video streams captured by VLC in .ts format into .mp4 format, using either HandBrake or 1-Click Video Converter. Both take advantage of hyper threading.

HandBrake easily pushes my i7 Mini to just below 100 degrees and over 1,000% CPU leaving less than 10% idle. The i7 MM case becomes very hot to the touch. 1-Click Converter doesn’t use quite as much CPU but the case still gets too hot to leave your hand on.

Meanwhile my 2014 MM running 1-Click Video Converter runs cool to the touch even when the 2 cores are maxed out. Obviously the i7’s 6 cores running at 3.2 GHz should generate more heat than the i5’s 2 cores running at 2.6 GHz. But the 2018 6 core i7 MM runs hotter to the touch than the 2014 dual core i5 even when both are stilling idle AND both reading roughly 60 degrees. So maybe the 2018 mini does a better job of using the case as a heat sink????

GetRealBro

 

[reks2004]

Hmmm i have strange situation, as we can see on the attached picture, i have REQ 3,2GHZ but Max is only 2,7 and AVG 2,6. I can reach 3Ghz and doesn't matter if i use only one or more core.

Any idea? (i have xeon E5)

 

[HDFan]

as we can see on the attached picture

????

 

[reks2004]

Added

 

[mrkapqa]

if someone has an older version that still functions with mac os snow leopard or lion - please share!