Survey: Only 5.6 Percent of Ryzen 9 3900X Hit Advertised Speeds, Most Other Models Suffer, Too

Overclocker and hardware reviewer Der8auer, widely known for his Intel delidding tools and overclocking videos, has released the results of a survey he conducted late last month concerning Ryzen 3000's ability to reach its advertised boost clocks. Only 5.6% of respondents reported that their Ryzen 9 3900X is reaching its rated boost speed. The results are somewhat better with other SKUs, but still indicate that the majority of Ryzen 3000 series processors are not hitting their rated boost speeds. 

Users and reviewers alike have been questioning whether or not AMD's new CPUs are always able to boost to the advertised clock speeds. We recently published an analysis on the 3600X detailing Ryzen 3000's new boosting behavior, and AMD confirmed that only one core on any given CPU is guaranteed to hit the rated boost clock. However, according to the survey, more users aren't even reaching the advertised frequency on any core.

Der8auer's survey obtained the performance data of 2,700 systems from users who were asked to run the single threaded benchmark on Cinebench R15 and record the maximum clock speed using HWInfo (which was recommended by AMD). Most users reported that they were not able to hit the advertised boost clock, though many were within 25 MHz.

(Image credit: YouTube / Der8auer)

At AMD's best, about half of Ryzen 5 3600 users reported their CPU was boosting correctly, and at worst, only 5.6% of Ryzen 9 3900X users reported that their CPU was boosting correctly. Most users were within 100 MHz of the advertised boost clock, but there was still a significant number who were more than 100 MHz away.

Der8auer does make it clear that the survey was not perfectly scientific, however. Firstly, not all users used the exact same hardware, but that is to be expected, and Der8auer says he went through every single result over three days to make sure BIOS version, AGESA version, and everything else was consistent and labeled. He also discarded some results: outliers, systems using unusual setups like chillers, and users who reported that they used PBO.

He does admit that users who weren't getting the rated boost clocks would be more likely to submit their result than users who had no issues, something which could skew results, and that he could not ensure whether or not users applied the Windows 10 update that ensures the Windows scheduler would be using the fastest core for single-threaded workloads.

On the other hand, though, the data more or less demonstrates that most users are not getting the experience promised by AMD and Der8auer says if a specific Windows version or something is required to achieve the rated boost, AMD should make that clear to its users.

There also doesn't seem to be a trend of certain motherboards and certain BIOSes being able to boost more reliably than others, something that earlier testing from reviewers like Hardware Unboxed thought may be the case when testing a much smaller sample of motherboards and just one 3900X. Der8auer states there might be something to Hardware Unboxed's findings, but also that it's not quite as simple as picking a motherboard and getting the clock speeds AMD promised.

Despite the controversy over AMD's advertised boost clocks, Der8auer still says he recommends all the Ryzen 3000 CPUs; however, he also states that the results in his survey were much worse than he expected and is worried about whether or not AMD can solve this issue in a timely manner. He concludes his findings wondering why AMD would advertise these clock speeds.

"Why did AMD feel that it is necessary to advertise the boost or give the people false expectations and false hope for something that they cannot get? Why did they have to do the 3900X at 4.6 [GHz] when they probably clearly know that most of those CPUs would never maintain this speed? It was clearly never necessary to do this; it's completely unnecessary. The CPUs are good enough the way they are. They deliver, and they're good. But those frequency values are just completely wrong." - Der8auer

AMD hasn't yet made a statement on whether or not this is the intended behavior of Ryzen 3000 CPUs, or if there is a fix in the works.

Matthew Connatser

Matthew Connatser is a freelancing writer for Tom's Hardware US. He writes articles about CPUs, GPUs, SSDs, and computers in general.

  • mdd1963
    Now would be a good time to show the few seconds of video where the one AMD guy scribbling on a screen shows a graph showing that PBO might give 4.75 GHz with good cooling... :)
    Reply
  • AlistairAB
    I've built many Ryzen 3000 systems and all of them will hit the boost clock, while under low utilization, confirmed by a look at a recorded HWInfo max clock speed reading. Doesn't mean you'll get the boost clock under a full single threaded Cinebench run. Why is that conveniently the definition now?

    My 8086k doesn't hit 5ghz under Cinebench ST runs either. Same story. The 9900k hits it because unlike the 8086k the boost is for 2 cores, not 1. And the 9900k will throttle itself under full load also, and not hit its all core boost speed either. A little perspective is necessary.

    This is like only being mad at LG for 1ms lies/misleading claims, but not being against Asus or TN panel makers doing the same thing.
    Reply
  • NightHawkRMX
    Really, boost clocks are confusing.

    Some CPUs hit their rated turbo under a 1 core workload, others under a 2 core workload. Some will only hit that speed when really cool, and some are duration limited.

    I dont really see any 9900k hitting at 5ghz either from the factory. Its like 4.8/9
    Reply
  • TCA_ChinChin
    Shame that not many Ryzen 3rd gen live up to their advertised clock speeds. Probably would have been better for AMD just to lower the advertised boost clocks by like 100 MHz and it probably wouldn't have been such a big deal.
    Reply
  • bit_user
    AlistairAB said:
    I've built many Ryzen 3000 systems and all of them will hit the boost clock, while under low utilization, confirmed by a look at a recorded HWInfo max clock speed reading. Doesn't mean you'll get the boost clock under a full single threaded Cinebench run. Why is that conveniently the definition now?
    The way the article is worded, it sounds like HWInfo is sampling the clock and they're just reporting the max value. Normally, you'd expect that to be at the start of the run. If so, that seems fair to me - not looking at sustained or final clocks, but just the max value that it reaches at any point.

    That said, you have a point that maybe the max boost clock assumes a less parallel, non-vectorized workload. AMD has said Ryzen 3k doesn't have any explicit AVX2-downclock, but that it's just subordinate to their normal power & thermal management. That's not the same as saying those workloads won't affect your clock speeds, however. Maybe they should rerun the experiment using a workload like single-threaded Javascript parsing.

    Anyway, it sounds like he also collected data on cooling solutions (i.e. if he's able do discard results using chillers). I wonder how much the situation improves, if you exclude everyone using the boxed cooler.
    Reply
  • alextheblue
    bit_user said:
    AMD has said Ryzen 3k doesn't have any explicit AVX2-downclock, but that it's just subordinate to their normal power & thermal management. That's not the same as saying those workloads won't affect your clock speeds, however.
    This was my immediate thought. Workload type affects Intel boost too... the whole affair just kind of makes me chuckle.
    Reply
  • Deadhound
    Stupid survey. Doesn't account for any kind of cooling, or people just trolling the survey. Also doesn't account for the possiblity of bad boards/BIOS and how the newer boards might actually be shittier than old cheap boards:
    https://www.igorslab.media/en/three-x570-motherboards-tested-wrong-sensor-values-faulty-precision-boost-overdrive-pbo-and-different-performance/2/
    Was a dude on AMD-subreddit claiming he made 10+ fakes to make AMD look bad.
    Reply
  • TerryLaze
    AlistairAB said:
    My 8086k doesn't hit 5ghz under Cinebench ST runs either. Same story. The 9900k hits it because unlike the 8086k the boost is for 2 cores, not 1. And the 9900k will throttle itself under full load also, and not hit its all core boost speed either. A little perspective is necessary.
    It's not the same story,for intel if you go with a high performance mobo it will make the CPU boost to it's max clocks...and then some,the whole story about the 9900k being a power hog and being very hot comes from this fact.
    You can also look at silicon lottery,100% of 9900k hit 4.8Ghz all core on normal workloads.
    100% of the 9900k also hits 4.6 on all core for AVX while for the 3900x it's only 4Ghz for all core.
    https://siliconlottery.com/pages/statistics
    Reply
  • hotaru251
    My question is will this lead to a lawsuit?

    we see nem sued (and lose) due to not true core cpu's....if ppl expect these speeds and cant get them wouldnt this be false advertisement and not selling what buyer is promised?
    Reply
  • Lunetouche
    He discounted a large number of results that were just 25mhz under max boost. That's less than .5% which could easily be a measuring error or max missed due to the rapid changes evading the polling. Personally my 3800x is boosting fine on all cores, the trick being keeping it nice and chilly. 360mm AIO and as many fans as I could cram in my 011 dynamic :p you know what never hit full boost ? My dud of a 5930k :p
    Reply