Guru3D.com
  • HOME
  • NEWS
    • Channels
    • Archive
  • DOWNLOADS
    • New Downloads
    • Categories
    • Archive
  • GAME REVIEWS
  • ARTICLES
    • Rig of the Month
    • Join ROTM
    • PC Buyers Guide
    • Guru3D VGA Charts
    • Editorials
    • Dated content
  • HARDWARE REVIEWS
    • Videocards
    • Processors
    • Audio
    • Motherboards
    • Memory and Flash
    • SSD Storage
    • Chassis
    • Media Players
    • Power Supply
    • Laptop and Mobile
    • Smartphone
    • Networking
    • Keyboard Mouse
    • Cooling
    • Search articles
    • Knowledgebase
    • More Categories
  • FORUMS
  • NEWSLETTER
  • CONTACT

New Reviews
Hitman III: PC graphics perf benchmark review
TeamGroup CX2 1TB SATA3 SSD review
EVGA GeForce RTX 3070 FTW3 Ultra review
Corsair 5000D PC Chassis Review
NZXT Kraken X63 RGB Review
ASUS Radeon RX 6900 XT STRIX OC LC Review
TerraMaster F5-221 NAS Review
MSI Radeon RX 6800 XT Gaming X TRIO Review
Sapphire Radeon RX 6800 NITRO+ review
Corsair HS70 Bluetooth Headset Review

New Downloads
CrystalDiskInfo 8.10.0 Download
SiSoft Sandra 20/20 download v30.92
AMD Radeon Adrenalin Edition 21.1.1 driver download
CPU-Z download v1.95
Intel HD graphics Driver Download Version: DCH 27.20.100.9168
HWiNFO Download v6.41 (4355 Beta)
GeForce 461.33 hotfix driver download
Prime95 download version 30.4 build 7
AIDA64 Download Version 6.32.5620 beta
3DMark Download v2.16.7117 + Time Spy


New Forum Topics
2TB version Samsung 980 Pro with 136 Layer (V-NAND v6) surfaces in webshops 3080 Owner's thread! AMD Ryzen 2021-2022 roadmap with codenames leak - Van Gogh and Warhol GeForce Hotfix Driver Version 461.33 Kozary's Triple SLI Thread Gabe Newell talks about Cyberpunk 2077 and sympathizes with CD Projekt Intel is satisfied about 7nm progress Thermaltake ToughDesk 500L RGB Is a Battlestation Gaming Desk Quick question: newer gen GPU for 600W PSU Review: Hitman III: PC graphics performance benchmark analysis




Guru3D.com » News » Der8auer: "Good number of Ryzen 3000 chips does not reach advertised boost clock"

Der8auer: "Good number of Ryzen 3000 chips does not reach advertised boost clock"

by Hilbert Hagedoorn on: 09/02/2019 04:58 PM | source: Der8auer | 143 comment(s)
Der8auer:

A few weeks or so we reported that the AMD Ryzen 3000 series boost clocks can vary quite a bit, even per motherboard manufacturer. In that same timeframe overclocker Der8auer requested users to send in data, basically collecting what boost clocks they achieve on a single thread.

The results are in, it seems specifically a 5.5% for the Ryzen 9 3900X processors jumps stay in the logical spectrum e.g. does did reach the advertised single thread boost speed. To determine the maximum boost clock, everybody that joined in has been asked to run single-threaded Cinebench R15 and then verify the clock speed using the HWInfo tool. Herein already a bit of an issue can be found, as running the two together can already spawn multiple threads on load.

In total, a number of 2,700 processors have been tested and results collected and yes, 5.5% of the Ryzen 9 3900X achieved the advertised boost clock of 4.6 GHz. The majority of procs stay within 4.5 and 4.75 GHz.  You should also be aware that 68% of the test results indicate a boost of 4.5 GHz or better, however, 32% did not. And that last bit I would agree, is weird. 

 

 

Are the results reliable?

While the test itself is pretty self-explanatory, some questions do arise. The test are consumer-based, ergo Der8auer has had no control over the test group. Simple things can influence that data; type of cooling used (which matters for Ryzen), neither is there an overview of how the motherboards are configured, or even if the advised Windows Energy settings mode and AMD chipset software is installed. Also, an inexperienced end-user might forget that applications run in the background and always need tyo be halted - if they did not do so it might have triggering other threads and thus lowering clocks. Also, in the weeks that Der8auer (Germany based) asked for the test results, there was a heatwave here in the EU, all these small factors can and will have an effect on performance one way or the other. Last but not least, probably some folks at intel read about the request and entered data as well? Who knows right? That's the problem with public data.

So are the results reliable? Well, it's a pretty good indication but not 100% reliable due to the aforementioned reasons. Then again, the number of entries make the results certainly statistically significant. The test does show there is something going on alright, the chips should be able to boost to its maximum advertised frequency (or at least close towards it in the 4.5+ GHz realm.).

Does this affect your performance?

No, all measurements to date are valid, if there is something going on, then the perf is the same as previously. If a chip is slower, then it also was that during any testing. It works vice versa, results could only become faster. Ryzen 3000 is a complicated processor. AMD already has explained in-depth that there are many variables in play that determine the Turbo single thread bin. The right thread also needs to be prioritized towards the fastest / best core, as not all cores are equal.



Der8auer: Der8auer:




« EK Offers EK-Momentum MSI Z390 MEG Godlike D-RGB Monoblock · Der8auer: "Good number of Ryzen 3000 chips does not reach advertised boost clock" · Homeworld 3 announced for PC »

Related Stories

Der8auer LN2 Clocks 18-core Core i9 7980XE To 6 GHz (Uses 1000 Watts Energy) - 09/25/2017 02:06 PM
Pro-overclocker Der8auer did his trick on the new 18-core Core i9 7980XE. He applied LN2 cooling, and was able to reach a tweak to 6 GHz, that however did results into a power-consumption of a scary ...

Der8auer delids Intel 12 and/or 18 core Skylake-X CPUs - 09/05/2017 05:49 PM
Pro overclocking guru Der8auer retreived early samples of Intel upcoming 12-18 core i9 Skylake-X series CPUs. Hedecided to use the Delid-Die-Mate-X delidding tool and succesfully delidded at least one...

Pro Overclocker Der8auer Feels X299 is a Platform Disaster for Overclocking - 06/29/2017 07:38 AM
As you have been able to notice from my revies, the X299 platform has massive challenges in overclocking (and other segments as well). Pro-overclocker Der8auer now states that the X299 Platform a &amp...

Der8auer Delid-Die-Mate 2 Released - 01/19/2017 09:36 AM
Fancy a bit more cooling on your Intel processor? Brave enough to try delidding but don't know how to ? Well, der8auer designed revision two of his easy removal ("delidding") tool of tha...


29 pages 1 2 3 4 > »


barbacot
Senior Member



Posts: 506
Joined: 2007-09-24

#5707023 Posted on: 09/02/2019 05:43 PM
As a user, you shouldn't really mind the slightly lower boost clock. If it is slightly lower, lets say 100~150Mhz.
But, as a consumer, you should be furious. If it is one of your key selling points, you must provide it at 100%.
I like AMD, I own a Ryzen 7 1700 and an RX 480, and that's because -at least when it comes to pricing and effort- they are more respectful towards the consumer compared to nVidia and/or Intel.
Hell, just a few hours ago I ordered my RX 5700 XT. But if the "downgrade to provide longevity" thing was planned beforehand -though nothing factual points to this for now-, it's an outright scam and I'll reconsider future purchases.

It is not something new or original - the same scam (marketing strategy) is used for a long time in the auto industry: when you want to buy a new car they advertise a fuel consumption that is just unreal sometimes and you see when you drive the car that you can never reach that fuel economy no matter how economically you drive and then you find out that the advertised rate is obtained in "certain" conditions that can never be replicated in the real world...

fry178
Senior Member



Posts: 1656
Joined: 2012-04-30

#5707025 Posted on: 09/02/2019 05:44 PM
i think they just were a little to optimistic about expected avg clocks/perf, and reality said "nope".
then again, its called boost, not minimum clock, and disclaimers always say "up to",
no matter what brand anyway.
i now look at base clocks being all core/full load freq, and boost clocks as a ST frequ.

then again running 3600 a 240 hydro, all auto oc/pbo etc turned off in bios, power/settings on normal/reg/auto,
i get 3900-4000 MT, and 4075-4100 on ST.

@
Mpampis

please explain why YOU would reconsider brands, when you dont even own 3xxx,
nor have experienced that "issue"?

sure, past 10% its not irrelevant nor statistical error, but as hilber said,
too many variables involved that are not under control of the (data) collector...

Neo Cyrus
Senior Member



Posts: 9607
Joined: 2006-02-14

#5707027 Posted on: 09/02/2019 05:47 PM


Well I used HWInfo to monitor as was instructed, and according to that on a 3900X with stock settings I get a max boost of 4.5GHz, never 4.6GHz, even if it is for a second. According to HWInfo my chip might be a Qualification Sample? HWInfo lists it as an Engineering Sample, but the in its serial says QS. So is HWI wrong, or...?

That max boost of 4.5GHz scores 201 in Cinebench unless I have all background junk closed, in which case it scored 207. In that case I might as well just leave it at an all-core OC of 4325MHz which I found to be seemingly stable at 1.39V, sure beats the 1.48-1.488V the auto settings pump through the chip. That also scores 201 single core with tons of stuff in my background.

Side note: My board goes nuts and doesn't allow actually decoupling the Infinity Fabric clock from the memory clock. Setting them separately results in half (of RAM) Infinity Fabric speed no matter what. All settings must be auto and at 3600MHz or lower, then the IF will run at the proper 1800MHz to match the RAM... aaaaand my RAM is stuck on a 2T command rate for some reason. All this, despite AMD themselves confirming Zen 2 is designed to handle an IF clock with a 1:1 ratio to the RAM up to 3733MHz. Good job MSI, you mouth breathing proto chimps.

This has been the messiest CPU upgrade I've ever had, great performance, but just figuring out what the hell is going on has been a convoluted clown fiesta.

fry178
Senior Member



Posts: 1656
Joined: 2012-04-30

#5707035 Posted on: 09/02/2019 05:56 PM
never saw anyone mentioning numbers higher than 1800 for IF (3600 on ram) when 1:1.
past that it, will switch to 2:1, which is normal.
3533 might most of the time even work better, as boards use auto on some settings (not man).

gerardfraser
Senior Member



Posts: 3345
Joined: 2008-03-08

#5707037 Posted on: 09/02/2019 06:04 PM
I have a Chip that can run past Max boost with Beta BIOS,I am running the newest BIOS with lower boost but the strange thing is.I get all the stability with the newest BIOS and the same performance.So I am good.

Not a good look for AMD at all and it amazing how they can mess this easy stuff up.
People need to get past the number thing though.AMD clock for clock is matching or passing Intel as of today in applications.Intel is faster 95% of games by a little bit not even worth talking about and I am sure if developers designed games for AMD in mind as much as Intel,that would be different.

The AMD CPU'S are still fast little buggers despite AMD clock reporting in third party applications or there own Ryzen Master.
I will keep building AMD systems until Intel can come up with something better,then Intel all the way.

never saw anyone mentioning numbers higher than 1800 for IF (3600 on ram) when 1:1.
past that it, will switch to 2:1, which is normal.
3533 might most of the time even work better, as boards use auto on some settings (not man).

Ryzen 3600X I run 3800 The memory clock (mclk), the memory controller clock (uclk), and the infinity fabric clock (fclk)



29 pages 1 2 3 4 > »


Post New Comment
Click here to post a comment for this news story on the message forum.


Guru3D.com © 2021