Power Consumption

The nature of reporting processor power consumption has become, in part, a dystopian nightmare. Historically the peak power consumption of a processor, as purchased, is given by its Thermal Design Power (TDP, or PL1). For many markets, such as embedded processors, that value of TDP still signifies the peak power consumption. For the processors we test at AnandTech, either desktop, notebook, or enterprise, this is not always the case.

Modern high performance processors implement a feature called Turbo. This allows, usually for a limited time, a processor to go beyond its rated frequency. Exactly how far the processor goes depends on a few factors, such as the Turbo Power Limit (PL2), whether the peak frequency is hard coded, the thermals, and the power delivery. Turbo can sometimes be very aggressive, allowing power values 2.5x above the rated TDP.

AMD and Intel have different definitions for TDP, but are broadly speaking applied the same. The difference comes to turbo modes, turbo limits, turbo budgets, and how the processors manage that power balance. These topics are 10000-12000 word articles in their own right, and we’ve got a few articles worth reading on the topic.

In simple terms, processor manufacturers only ever guarantee two values which are tied together - when all cores are running at base frequency, the processor should be running at or below the TDP rating. All turbo modes and power modes above that are not covered by warranty. Intel kind of screwed this up with the Tiger Lake launch in September 2020, by refusing to define a TDP rating for its new processors, instead going for a range. Obfuscation like this is a frustrating endeavor for press and end-users alike.

However, for our tests in this review, we measure the power consumption of the processor in a variety of different scenarios. These include full peak AVX workflows, a loaded rendered test, and others as appropriate. These tests are done as comparative models. We also note the peak power recorded in any of our tests.

First up is our loaded rendered test, designed to peak out at max power.

In this test the 3995WX with only 64 threads actually uses slightly less power, given that one thread per core doesn’t keep everything active. Despite this, the 64C/64T benchmark result is ~16000 points, compared to ~12600 points when all 128 threads are enabled. Also in this chart we see that the 3955WX with only sixteen cores hovers around the 212W mark.

The second test is from y-Cruncher, which is our AVX2/AVX512 workload. This also has some memory requirements, which can lead to periodic cycling with systems that have lower memory bandwidth per core options.

Both of the 3995WX configurations perform similarly, while the 3975WX has more variability as it requests data from memory causing the cores to idle slightly. The 3955WX peaks around 250W this time.

For peak power, we report the highest value observed from any of our benchmark tests.

(0-0) Peak Power

As with most AMD processors, there is a total package power tracking value, and for Threadripper Pro that is the same as the TDP at 280 W. I have included the AVX2 values here for the Intel processors, however at AVX512 these will turbo to 296 W (i9-11900K) and 291 W (W-3175X).

AMD TR Pro Review: 3995WX, 3975WX, 3955WX CPU Tests: Rendering
Comments Locked


View All Comments

  • Threska - Friday, July 16, 2021 - link

    One could have more than one.


    Especially depending upon the memory type.

  • croc - Monday, July 19, 2021 - link

    The first year anniversary of the Threadripper Pro! And a timely review to celebrate it! Not the CPU, the release... Because that might be the last Threadripper we see for quite some time....

    I get the impression that AMD has been in the position of runer up for so long now that they don't know how to capitalize on a lead. Either that or by selling off their fab they have lost touch with how to design for fabrication, which might explain their failure to deliver product at the 7NM node. Unfilled Epycs, no Zen 3 Threadrippers...
  • mode_13h - Monday, July 19, 2021 - link

    What a croc!
    ; )

    > that might be the last Threadripper we see for quite some time....

    The announcement for Chagall is rumored to be coming in August, with volume shipping in September. Have you heard otherwise?

    > by selling off their fab they have lost touch with how to design for fabrication

    Huh? Why do you think the issue is design-related?

    And if we're contemplating counter-factuals, then let's not lose sight of the fact that GF is still on 12 nm. And if AMD still owned them, we should also consider whether the whole enterprise would still be doing business, at all.

    > which might explain their failure to deliver product at the 7NM node.

    Or maybe they're in the same boat as everyone else, facing higher demand and restricted supply?
  • croc - Tuesday, July 20, 2021 - link

    Lemmee see.... Wasn't Chagall an artist? And wasn't the code name for the Threadripper Zen 3 to be Genesis Peak? And aren't you quoting a rumor? Have there been ANY leaked benchmarks? Has ANY processor EVER launched without leaked benchmarks?

    The lack of things often speak volumes...
  • mode_13h - Tuesday, July 20, 2021 - link

    > And aren't you quoting a rumor?

    Yes, that's exactly what I said.

    > Have there been ANY leaked benchmarks?
    > Has ANY processor EVER launched without leaked benchmarks?

    It's a niche product. There typically aren't many motherboard options for them (and even fewer, at launch). I think it's not surprising, if there aren't any benchmark leaks as of yet. There won't be many engineering samples floating around.

    Anyway, we'll know within about 5 weeks if there's any truth to the rumor.
  • croc - Wednesday, July 21, 2021 - link

    I will know when a bios with AGESA support for a Zen 3 Threadripper arrives. None have arrived yet, and typically they do so about a month prior. Chagal is an artist, and artist names are usually reserved for apus. Supposedly the next TR will use the TRX40 socket, which means that there are many MBs for testing / leaking. Given that the Epyc still uses the SP3 socket, I believe this to be tue The latest 'chagall' rumor is that it won't release before Sept, mebbe as late as November. Gotta love them movable goal posts

    Personaly, I don't think that AMD can get the cores to deliver at the frequencies required for an HEDT product Hell, given their back orders for their server chip I doubt that they even care. Bigger fish to fry, all that. Still, a bit of egg-on-face for Dr. Su. And niche? don't tell the CGI world that they are niche...
  • Qasar - Wednesday, July 21, 2021 - link

    sorry croc, but you could be wrong, looks like TR based on zen 3 could be Chagall, and on both STRX4 and SWRX8, at least according to here :

    from moores law is dead, a video about TR zen 3 :
    but no concrete info as of yet, but a search on google for threadripper zen 3, all seems to say the same, code name chagall
  • croc - Saturday, July 24, 2021 - link

    No new BIOS since April... 7 days to August, so looks like another no-show, eh...
  • Qasar - Saturday, July 24, 2021 - link

    and the point is ? considering their has been no official announcement from amd when zen 3 TR is to be released, why would there be a bios for it yet ?
    the point of my reply was that you could be wrong on the code name, not the release date.

    at least amd HAS a HEDT cpu, when was the last one from intel ?

    bottom line : all there are, are rumors, which should be taken with some salt.
  • croc - Monday, July 26, 2021 - link

    That would be the Xeon w-3175x

Log in

Don't have an account? Sign up now