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

Page 5 - 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
660
1,003
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:

Det0x

Golden Member
Sep 11, 2014
1,027
2,953
136
Alrighty, have finally figured out my problems after a windows reinstall :)

Zen4 7950x @ 105 cTDP = 122w average power --> ~38.7k Cinebench r23 MT score
  • ST performance efficiently score = ~139
  • MT performance efficiently score = ~11599
1665607055843.png

Zen4 7950x @ 65 cTDP = 80w average power --> ~33.5k Cinebench r23 MT score
  • ST performance efficiently score = ~140
  • MT performance efficiently score = ~14202
1665607141679.png

Zen4 7950x @ 50w PPT limit = 50w average power --> ~24.8k Cinebench r23 MT score
  • ST performance efficiently score = ~143
  • MT performance efficiently score = ~14076
1665607244571.png

Zen4 7950x @ 35w PPT limit = 35w average power --> ~18.5k Cinebench r23 MT score
  • ST performance efficiently score = ~188
  • MT performance efficiently score = ~11525
1665607389595.png
 
Last edited:

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
Zen4 7950x @ 65 cTDP = 80w average power --> ~33.5k Cinebench r23 MT score
  • ST performance efficiently score = ~140
  • MT performance efficiently score = ~14202
1665607141679.png
That MT efficiency at 65W TDP is insane. o_O

ST efficiency increasing at lower TDP shows that it's constantly boosting well beyond the efficiency infliction point. I guess only a fixed frequency could avoid that. (Or possibly a low temperature limit?)
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
That MT efficiency at 65W TDP is insane. o_O

ST efficiency increasing at lower TDP shows that it's constantly boosting well beyond the efficiency infliction point. I guess only a fixed frequency could avoid that. (Or possibly a low temperature limit?)
For an ASRock board/bios, all that I can find is if you use the max temp feature. No TDP setting or ECO mode that I can find (or anybody has told me about) Tight now I an running 90c, but might lower that later. I would LOVE to have that TDP feature ! It did lower the wattage, and I see no difference in times, but I am still at 35k on cb23.
 

Det0x

Golden Member
Sep 11, 2014
1,027
2,953
136
For an ASRock board/bios, all that I can find is if you use the max temp feature. No TDP setting or ECO mode that I can find (or anybody has told me about) Tight now I an running 90c, but might lower that later. I would LOVE to have that TDP feature ! It did lower the wattage, and I see no difference in times, but I am still at 35k on cb23.
You should have something like this in/around your PBO settings
1665610587435.png

If you have no eco mode you can set manually PBO PPT limit to 88w
1665610638216.png

If you don't have any PBO settings in your ASRock part of the bios, you should find this in the AMD part of the bios (advanced)
1665610707512.png

Hope this helps :)
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
You should have something like this in/around your PBO settings
View attachment 69008

If you have no eco mode you can set manually PBO PPT limit to 88w
View attachment 69009

If you don't have any PBO settings in your ASRock part of the bios, you should find this in the AMD part of the bios (advanced)
View attachment 69010

Hope this helps :)
Not on my screen... It was only the beta bios even had the throttle limit. I am on the latest beta bios.
20221012_135956[1].jpg
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
Quite good efficency at 65W but when you do a compromise between pure perf and perf/watt then it s obvious that the 7950X really stretch its legs within a 88-105W range, at 65W the 7900X does almost as well despite the core count deficit at 87% of the 7950X@65W MT perfs.

Computerbase added a few numbers for the 7900X before ending its tests :


 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
at 65W the 7900X does almost as well despite the core count deficit.
Ignoring the uncore overhead (so these will likely be worse in reality) 7900X should achieve a similar MT efficiency at around 49W, likewise 7700X at around 33W and 7600X at around 24W, lol...
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
I found a setting that I like ! I tried PBO and 170 watt, and somehow when I booted it was very slow and said 0 is the PPT. So on a different screen I found something that said offset, (negative) and I set that to 25. When I booted and I run cb23, it says 142 watt and 13 watt SOC, so less than 170 ! And the temp at 90. cb23 still running, I will update when it finishes !

Edit: First run 35,436 . So it did not affect performance at all and uses the same power as a 5950x now !

I am happy. This is what I wanted when Zen 4 came out.
 
Last edited:

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,629
136
I found a setting that I like ! I tried PBO and 170 watt, and somehow when I booted it was very slow and said 0 is the PPT. So on a different screen I found something that said offset, (negative) and I set that to 25. When I booted and I run cb23, it says 142 watt and 13 watt SOC, so less than 170 ! And the temp at 90. cb23 still running, I will update when it finishes !

Edit: First run 35k and change. So it did not affect performance at all and uses the same as a 5950x now !

A -25 offset is pretty large across all cores. It might not be really stable with that large an offset but who knows, maybe you got a top chip.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
A -25 offset is pretty large across all cores. It might not be really stable with that large an offset but who knows, maybe you got a top chip.
Well, after prliminary tests, it looks fine, so I rebooted linux, and now am running 30 WCG tasks and the 3090TI is running F@H, so its maxed. We will see how it does.
 

.vodka

Golden Member
Dec 5, 2014
1,203
1,537
136
a0qfn2Z.png


3900X + PBO fmax enhancer (ASUS' official EDC bug for Zen2), stock PPT TDC EDC (142w 95A 140A), 1900MHz fclk / 1900MHz uclk / 3800MHz mclk so the I/O die eats some more into the total power budget than usual.

Really nice to see efficiency going up with newer generations. Zen4 in particular is insane. Good stuff!

----------------

edit: I was curious what would happen if I gave it a higher power budget. So, 180W PPT (~20w for the I/O die plus some headroom for the cores), 110A TDC, 140A EDC (can't go over 140A on AGESA v2 >1.2.0.3c or else you get a 1.42v voltage cap)

JjLWOJ7.png


It's become more efficient at least on this workload, lol
 
Last edited:

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
I just realized. My new 7950 at 142 watts, is it on your chart ? I normally let it run linux, but I could install office and try it at 142 watts, but didn't detox do this at all sorts of wattage settings with his 7950x ?
 

BorisTheBlade82

Senior member
May 1, 2020
660
1,003
106
So after having been some days off, quite some catching up to do...

@Det0x
These are some quite impressive results. Your 65w cTDP numbers are a bit better than the ones I got from Andreas Schilling - so you might have a Golden Sample ;) Aside from changing the cTDP did you tune something else as well? What kind of RAM do you have (although CB23 should not be very sensitive in this regard)?

I just realized. My new 7950 at 142 watts, is it on your chart ? I normally let it run linux, but I could install office and try it at 142 watts, but didn't detox do this at all sorts of wattage settings with his 7950x ?
No, but I will add the 105w and 65w cTDP numbers of @Det0x shortly. In the past I tried to only include stock results in order not to clutter the rankings. But I have a feeling that with this generation from both vendors the interest regarding performance and efficiency under limited TDPs has been growing lately.
So I will also re-add the 12900K@125w TDP for the sake of comparison.

/edit:
@Det0x
Would you mind doing another run with stock settings? Your 105w TDP run was already faster than the only stock-run I have of the 7950X until now.
 
Last edited:
  • Like
Reactions: lightmanek

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
So after having been some days off, quite some catching up to do...

@Det0x
These are some quite impressive results. Your 65w cTDP numbers are a bit better than the ones I got from Andreas Schilling - so you might have a Golden Sample ;) Aside from changing the cTDP did you tune something else as well? What kind of RAM do you have (although CB23 should not be very sensitive in this regard)?


No, but I will add the 105w and 65w cTDP numbers of @Det0x shortly. In the past I tried to only include stock results in order not to clutter the rankings. But I have a feeling that with this generation from both vendors the interest regarding performance and efficiency under limited TDPs has been growing lately.
So I will also re-add the 12900K@125w TDP for the sake of comparison.

/edit:
@Det0x
Would you mind doing another run with stock settings? Your 105w TDP run was already faster than the only stock-run I have of the 7950X until now.
BTW, my results were from setting PBO(some sub-option) to -25 and max temp to 90. Its just so happens that Ryzen master now says 142 watt cpu and 13 watt SOC. You might want to put that in your notes.
 

BorisTheBlade82

Senior member
May 1, 2020
660
1,003
106
@Markfw
Yep, understood. That is why I refrain from putting them in the rankings. When we start chasing down that rabbit hole of max. temp settings, Vcore offsets and stuff, we will never see the light of day again - I hope you understand ;)

So I just updated the rankings and included the results from @Det0x with 65w and 105w cTDP as well as @.vodka 's 3900X (to my own surprise we did not have a sample of that before).
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
@Markfw
Yep, understood. That is why I refrain from putting them in the rankings. When we start chasing down that rabbit hole of max. temp settings, Vcore offsets and stuff, we will never see the light of day again - I hope you understand ;)

So I just updated the rankings and included the results from @Det0x with 65w and 105w cTDP as well as @.vodka 's 3900X (to my own surprise we did not have a sample of that before).
BUT, my motherboard has NO ECO setting. Obviously the bios are very different. But any sane person would not leave bios at stock for Alderlake, Zen4 or Raptorlake IMO. So we really need to figure out the best way to determine efficiency, and wattage at full load would seem to be it.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
@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.
 

Det0x

Golden Member
Sep 11, 2014
1,027
2,953
136
So after having been some days off, quite some catching up to do...

@Det0x
These are some quite impressive results. Your 65w cTDP numbers are a bit better than the ones I got from Andreas Schilling - so you might have a Golden Sample ;) Aside from changing the cTDP did you tune something else as well? What kind of RAM do you have (although CB23 should not be very sensitive in this regard)?
Runs were done with PBO CO enabled, handtuned for my cpu with golden CCD0 (SP121) and below average CCD1 (SP112).
Memory were running at 4800MT/s with all stock voltages and timings.

@Det0x
Would you mind doing another run with stock settings? Your 105w TDP run was already faster than the only stock-run I have of the 7950X until now.
Yes i can do a new run at "optimized bios settings" (all stock) in the next few days
 
  • Like
Reactions: BorisTheBlade82