Discussion PES | Assessing Power and Performance Efficiency of x86 CPU architectures

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
Dear Community,

so this is my first thread here as a long-time lurker - but I felt the desire to share a small hobby-project of mine from the last couple of months with you...

Performance Efficiency Suite - What is it about?
Most Reviewers solely focus on what they consider to be the most important aspect of modern CPUs - the absolute performance. But this is only one side of the equation. Today Power Efficiency is at least as important - or to be more precise: The amount of energy (Wattseconds or Joules) a CPU needs in order to accomplish a given workload. Sadly most Reviewers shy away from the extra mile it needs to assess this aspect. This suite measures the Total Package Power of a CPU while running the Cinebench R23 benchmarks first in single-threaded mode (1 run), then running in multi-threaded mode (for 10 minutes + whatever it takes to finish the last run). The results will be rendered in the provided Results.xlsx Excel file. To combine Efficiency and Performance there is also a score provided called Performance Efficiency Score (how amazingly inspired I am ;)).

In the meantime I was able to aggregate more than 80 samples from members of the 3DC & CB communities (see below).

How-To
  1. Unzip the latest release to wherever you want EXCEPT on your local OneDrive folder.
  2. Open Settings.txt and insert your local Cinebench23 Directory.
  3. Run PES Start - it will ask for Administrator rights as these are needed for measuring Package Power
  4. Wait until the Powershell finishes.
  5. Open the Excel file...
  6. Allow external connections (to the generated CSV-files with the data)
  7. Go to Data -> Refresh all
  8. Enjoy and share your results - just take a screenshot of what the Excel renders.
  9. If you want to do multiple measurements with different settings just copy the Excel file (inside the root-folder) before running and refreshing the data.

Some explanations about the Suite
  • This Suite has been made possible by Michael Möller and his amazing free and open-source Open Hardware Monitor and his .NET Library OpenHardwareMonitorLib.dll - Thanks a lot!!!
    Homepage: https://openhardwaremonitor.org/
    GitHub: https://github.com/openhardwaremonitor
  • The results for the Package Power look pretty accurate compared to the sparse data the internet provides. Seems, that the vendors are much more honest with those sensors than they are with temperature etc.
  • The suite basically consists some powershell scripts and an Excel file for presentation purposes
    • RunAsAdminWrapper.ps1
      This is needed to have a convenient relative path shortcut in the root folder and request admin-rights at the same time
    • Main.ps1
      • After setting up some stuff it basically starts the Cinebench R23 one at a time. It then checks for the "Cinebench.exe" process being active.
      • While this is true it queries the Package Power Sensor data with a lower bound of 10ms (in order to keep CPU-load of the script at bay).
      • After each run the aquired data gets pushed to CSV files located in the LogCsv subfolder.
    • Results.xslx
      • The Excel file basically just does some import, calculations and a hopefully nice presentation of the data.
      • Histogram
        The bold line shows a running average of the last 100 data-points which should be sufficiently accurate. The pale line shows each single data-point.
      • Calculation of Total Package Consumption
        To get that number we need the integral. That is why we first calculate the timeframe between two data-points and then multiply the measured value.
      • Everything else in that Excel is hopefully more or less self-explaining

Online Resources

Disclaimer
I am by no means a Powershell professional or a professional Reviewer. I was just sick of the lack of information and wanted to propose a low-effort solution. Any input for further improvement is highly welcomed. Please feel free to use/extend/rip-off this solution as you wish. But please share your findings to the world.
 
Last edited:

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
Thanks to @Kocicak we now already have results for Raptor Lake.

Some thoughts:
  • Performance Efficiency at ST has improved quite a bit compared to Alder Lake. And with the one exception of the R5 7600X it is also significantly ahead of Zen 4.
  • At MT Raptor Lake improves Performance Efficiency by around 70% compared to Alder Lake. It is at least able to be on par with the 5950X now, although still lagging quite a bit behind the 7950X - but at least Intel was able to reduce the gap a bit.
  • Taking into account that Intel7 is a full node behind TSMCs N5, this is not a bad showing - which in turn is good for consumers as heavy competition will help street-prices.
I have also updated all the matrices and rankings in posts #2 and #3.

@BorisTheBlade82
Would it be possible to insert an average lines on the chart for CPUs with multiple results? That way many possible parameters could be covered over time without adding more points for each single one.
Good point. I will try that when I find the time.
 

thigobr

Senior member
Sep 4, 2016
231
165
116
@thigobr
Your Zenbook seems to be limited to 12w PPT. If this is performance mode, how low is the limit for other modes?
This is quite good for energy consumption. But it also leaves a lot of performance on the table.

I don't know what happened there, probably a bug on the MyAsus app... I did the test again this time making sure performance mode was enabled. Sustained power is 25W in this mode

pes_25W.png
 

Cstops

Junior Member
Oct 14, 2022
5
17
41
I couldn't resist upgrading my 1800x to a 5700x (the price was right, had an itch to scratch, and it's a drop-in upgrade).

Here's the results I got. Same mobo (x370 Taichi, P7.1 BIOS) at stock settings (PBO defaults to 'auto') except for RAM (3,200 MT/s via XMP).

1666681273750.png

Relative to my old 1800x, looking at a 36% reduction in finish time in ST and and 30% per run in MT.
All while sipping power (-42% in ST and -54% in MT in measured "consumption," respectively) in this instance.
 
Last edited:

mmaenpaa

Member
Aug 4, 2009
78
138
106
AMD Ryzen 7 7700X BOX AM5 8C/16T 105W
ASROCK B650E PG RIPTIDE WIFI ATX MB AM5 (bios 1.11)
KINGSTON 32GB 6000MT/s DDR5 CL36 DIMM (@EXPO settings)
FRACTAL DESIGN Celsius S24 Water Cooling (manual setting)

All bios settings at default except EXPO memory (so PBO is *off*).

EDIT : After checking BorisTheBlade82 results I am thinking that Asrock AUTO profile has PBO enabled, as MT efficiency is so much lower than his results. I will try Ryzen master curve optimizer next.

1667647294169.png
 
Last edited:
  • Like
Reactions: lightmanek

mmaenpaa

Member
Aug 4, 2009
78
138
106
AMD Ryzen 7 7700X BOX AM5 8C/16T 105W
ASROCK B650E PG RIPTIDE WIFI ATX MB AM5 (bios 1.11)
KINGSTON 32GB 6000MT/s DDR5 CL36 DIMM (@EXPO settings)
FRACTAL DESIGN Celsius S24 Water Cooling (manual setting)
Windows 11 Pro

This is after Ryzen Master Curve Optimizer, (test 60 seconds). First try with 30 second test did not "take". Ie. PC would not boot fater applying & restarting. It seems that it is safer to use longer test to get stable result. Optimizing with 60 seconds took 1 hour and 25 minutes with 7700X.


1667670177503.png
 

JoeRambo

Golden Member
Jun 13, 2013
1,814
2,105
136
7950x @ 5Ghz static @ 1.025V, 2x32GB DDR5 6000C36, 16C32T, iGPU on
1670496354202.png


7950x @ 5Ghz static @ 0.97V, 2x32GB DDR5 6000C36, 16C16T, iGPU on
1670496409424.png


Not really "peak" efficiency runs, as SOC/IO/DDR5 voltages were high to support DDR5 speeds that don't really help CB23 at all and also integrated GPU was on, with dGPU it would be more efficient too.
And 5Ghz is not really a peak efficiency point either, but everyone knows that 5Ghz is almost 2x faster than 4.9Ghz. And obviously the cooler in mATX case was not up to task either.
 
Last edited:

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
@mmaenpaa
Thanks a lot for sharing these results of directly comparable SKUs of Cezanne and Rembrandt.
Just as expected there is no significant improvement for the latter in the CPU department - the results are a bit all over the place.
Interesting that Lenovo seem to have changed the power profiles quite significantly. 15w vs. 30w sustained - and Rembrandt seems to scale pretty well, being slightly on top in terms of performance efficiency at twice the wattage. Seems they wanted to unleash some performance in the newer generation.
Also, some interesting ST behaviour on battery - capped at 8w while MT is capped at 30w - do they cap the max. boost clock on battery?
I fully expect Phoenix Point to significantly improve at comparable wattages.
 

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
Release v0.8.0 - Let the Geeks out!
  • Geekbench 5 integration
  • You will now have the choice which test to run - so if you only want to run Geekbench 5, now you can (just takes around 3 minutes on an average machine).
  • Reworked Results Renderers (Results Excel.xlsx, Results LibreOffice.ods)
  • Reduced Cooldown between tests to 20 seconds
Some remarks about the Geekbench 5 integration
  • We measure the whole GB5 run - single-thread and multi-thread - together. It is not possible to run these two separately with the Tryout version. As both parts apply the same workload, this will very likely result in rather good scaling with ST-performance and efficiency of a CPU and rather bad with pure core-count. So GB5 should be considered a lightly-threaded workload.
  • While CB R23 is a sustained full load, GB5 is rather bursty. So the Boost behaviour of the CPU plays a significant role.
More Information: GitHub | Performance Efficiency Suite

Download v0.8.0

I hope you all will enjoy this and contribute some data.

/edit: Added GB5 Matrix and Rankings to the opening posts.
 
Last edited:

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
Geekbench 5 results from my HP Envy x13 AMD R7 4700U
On Battery
1672480169849.png

AC
1672480213159.png

As you can see, the boost behaviour of Renoir has a significant impact on Performance-Efficiency. For me, it is quite worth it - the negative performance impact on battery is not perceivable, while the extra efficiency is more than welcome and more than outweighs the former aspect.
 

Det0x

Golden Member
Sep 11, 2014
1,028
2,953
136
Always have so much problems when i try to run this software :cry:

When i click allow updating im getting this error
1673016898510.png

Then i get some of the data shown
1673016939520.png

All the data collected see to be there no ?

1673017240072.png
 
  • Like
Reactions: BorisTheBlade82

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
Always have so much problems when i try to run this software :cry:

When i click allow updating im getting this error
View attachment 74062

Then i get some of the data shown
View attachment 74063

All the data collected see to be there no ?

View attachment 74064
Did I already mention that I hate LibreOffice Calc?
Seriously. Could you just append the CSV files from the LogCsv subfolder as a Zip please? I will have a look or at least post your result.
 

Det0x

Golden Member
Sep 11, 2014
1,028
2,953
136
Did I already mention that I hate LibreOffice Calc?
Seriously. Could you just append the CSV files from the LogCsv subfolder as a Zip please? I will have a look or at least post your result.
Here you go, have saved the CSV files @ https://easyupload.io/m/5amdyp

7950x @ the following power targets:

260PPT
160PPT
105 cTDP
65 cTDP
50PPT

Please post best/interesting results :)
 
  • Like
Reactions: BorisTheBlade82

BorisTheBlade82

Senior member
May 1, 2020
663
1,014
106
Here you go, have saved the CSV files @ https://easyupload.io/m/5amdyp

7950x @ the following power targets:

260PPT
160PPT
105 cTDP
65 cTDP
50PPT

Please post best/interesting results :)

Okay, so here are your results. And as these are the first Geekbench 5 results we have from a person that is not me, I am quite surprised as well. To be honest, I would not have expected such bad performance-efficiency. In CB R23 a 7950x usually is better than my Renoir in multi-thread and more or less on par in single-thread.
Your system was in idle before testing and you also had no severe workloads running in parallel?
So beyond 105w there is almost no performance scaling while consumption goes up.

50w PPT
50w PPT.png

65w PPT

65w PPT.png

105w
105w PPT.png

160w

160w PPT.png

260w
260w PPT.png