Monday, September 2nd 2019

Der8auer: Only Small Percentage of 3rd Gen Ryzen CPUs Hit Their Advertised Speeds

World famous overclocker Der8auer published his survey of boost clocks found on 3rd generation Ryzen CPUs. Collecting data from almost 3,000 entries from people around the world, he has found out that a majority of the 3000 series Ryzen CPUs are not hitting their advertised boost speeds. Perhaps one of the worst results from the entire survey are for the 12-core Ryzen 9 3900X, for which only 5.6% of entries reported have managed to reach the boost speeds AMD advertises. However, the situation is better for lower-end SKUs, with about half of the Ryzen 5 3600 results showing that their CPU is boosting correctly and within advertised numbers.

Der8auer carefully selected the results that went into the survey, where he discarded any numbers that used either specialized cooling like water chillers, Precision Boost Overdrive - PBO or the results which were submitted by "fanboys" who wanted to game the result. Testing was purely scientific using Cinebench R15 and clock speeds were recorded using HWinfo (which got recommendation from AMD), so he could get as precise data as possible.
Der8auer comments that he still recommends Ryzen 3000 series CPUs, as they present a good value and have good performance to back. He just finds it very odd that AMD didn't specify what you need to reach the advertised boost speeds.

If you would like to see the more in depth testing, here is the English version of the video:

Add your own comment

248 Comments on Der8auer: Only Small Percentage of 3rd Gen Ryzen CPUs Hit Their Advertised Speeds

#201
Tatty_One
Senior Moder@tor
I have to say, I am not convinced citing non CPU hardware and other manufacturers "also" fail to meet (occasionally or a sustained?) boost speed is relevant and it certainly does not negate the topic of this thread, IMO any hardware that does not perform and generally sustain an advertised spec is simply wrong, I don't see any point in saying because something else also does not it kind of makes it OK ……… so can we change direction and stick to at least just CPU's please.
Posted on Reply
#202
Midland Dog
Midland Dog, post: 4109496, member: 168254"
well EVERY INTEL chip ive owned hit its turbo, if the sillicon cant hit those clocks then tell us so we dont expect intel crushing perf in older titles
heck even my old macbook pro (8,1) hit 3ghz for a minute before it hit 100c
Posted on Reply
#203
jaggerwild
I could have sworn I brought this up before (earlier threads but was ignored), I'm a sit back n watch now.
Posted on Reply
#204
Midland Dog
jaggerwild, post: 4109503, member: 61229"
I could have sworn I brought this up before(earlier threads but was ignored), I'm a sit back n watch now.
well then mate apply for an editorial role at tpu, you picked a topic of debate that keeps people at the site to either shitpost or take things to heart, all the marks of a good writer

jaggerwild, post: 4109503, member: 61229"
I could have sworn I brought this up before(earlier threads but was ignored), I'm a sit back n watch now.
since i spot an x99 owner, have you toyed with OCing ddr4? im more curious as to if the imc dictates the max mem clocks or if the type of dram matters, mainly because i have benched a ddr3 1333 kit at 2933, just curious thats all

Midland Dog, post: 4109504, member: 168254"
well then mate apply for an editorial role at tpu, you picked a topic of debate that keeps people at the site to either shitpost or take things to heart, all the marks of a good writer


since i spot an x99 owner, have you toyed with OCing ddr4? im more curious as to if the imc dictates the max mem clocks or if the type of dram matters, mainly because i have benched a ddr3 1333 kit at 2933, just curious thats all
on haswell
Posted on Reply
#205
jaggerwild
Nope, haven't really overclocked or even looked at my memory timings. I set at default, been working my ass off mostly. Hell that reminds me to check if the default V core is correct........UGH!! Back on topic, Intel cores ALL BOOST to said speed even when overclocked, AMD no only a few cherry cores will boost the others sit back n watch.
Posted on Reply
#206
Nordic
I just found this thread today, and it was amusing to see various people saying either user error, temperature constraints, or power constraints are the issue. I hope one day soon I get a bios update that allows my cpu to actually reach 4600mhz. I plan to install full custom watercooling hopefully later this year to raise my average boost clocks higher. These ryzen cpu's are really great.
Posted on Reply
#207
bogami
What you can expect from a waste die series! Not even close to GPU OC capability. 8 and less core - 25 gb/s of writing is easily surpassed by DDR3 on my system. 7nm technology is Marketing expression ! . That the truth is shrouded in so much lies that it hurts. they have not reached the six-year-old Intel technology. they touched in some places. Intel understands the word QUALITY. Think of how disappointed people are with so much inflating the waste series and still is being overpriced! Reviews also don't say much about the way it sells here. THIS AMD R 3000 Series is a proven sale of WASTE COREs ! I Wonder who from all the reviews pll has enough eggs to say it publicly !
Posted on Reply
#208
EarthDog
And what do you know... AMD sends out an email covering the boost clock issue...promising a FIRMWARE fix.

Now, if this really was an issue of cooling/power/board/silicon lottery/nominal conditions/user error/polling rates/AIB UEFI's, etc, they probably would have said so, right? Instead, they identified they have an issue and are correcting it via FW and not telling the client some line about "maximum" clocks and whatever other BS was brought up in this thread.

Here it is.......
[I]“AMD is pleased with the strong momentum of 3rd Gen AMD Ryzen™ processors in the PC enthusiast and gaming communities. We closely monitor community feedback on our products and understand that some 3rd Gen AMD Ryzen users are reporting boost clock speeds below the expected processor boost frequency. While processor boost frequency is dependent on many variables including workload, system design, and cooling solution, we have closely reviewed the feedback from our customers and have identified an issue in our firmware that reduces boost frequency in some situations. We are in the process of preparing a BIOS update for our motherboard partners that addresses that issue and includes additional boost performance optimizations. We will provide an update on September 10 to the community regarding the availability of the BIOS.”
[/I]
EDIT: Did anyone else notice they said "EXPECTED boost clock" and didn't try to split hairs on defining "maximum" or clarifying further what that meant?
Posted on Reply
#209
Darmok N Jalad
Midland Dog, post: 4109479, member: 168254"
except that for the last however many years boost clock was basically guaranteed, tell me an intel or amd gen prior to ryzen that couldnt hit its boost on a large scale. we arent just talking about half of the ryzens not hitting boost we are talking about ONLY 5.7% of a sku hits its boost, please tell me that you dont believe that only 5.7% of 9900k chips have hit 5ghz boost, because that is soooooo wrong
My W3690 has never officially hit the claimed boost of 3.73GHz. I don’t mind, because it runs at 3.6GHz all-core, and the base clock is 3.43GHz. It almost hit 3.7GHz on one core once. I’ve always viewed boost clocks as something you may get, but I am more concerned about the all-core sustained clocks, which are usually still higher than the rated base clock for all CPUs.
Posted on Reply
#210
ssdpro
At least the back and forth can stop. AMD finally admitted a firmware defect and is working on a patch.
“AMD is pleased with the strong momentum of 3rd Gen AMD Ryzen™ processors in the PC enthusiast and gaming communities. We closely monitor community feedback on our products and understand that some 3rd Gen AMD Ryzen users are reporting boost clock speeds below the expected processor boost frequency. While processor boost frequency is dependent on many variables including workload, system design, and cooling solution, we have closely reviewed the feedback from our customers and have identified an issue in our firmware that reduces boost frequency in some situations. We are in the process of preparing a BIOS update for our motherboard partners that addresses that issue and includes additional boost performance optimizations. We will provide an update on September 10 to the community regarding the availability of the BIOS.”
Posted on Reply
#211
EarthDog
ssdpro, post: 4109577, member: 131037"
At least the back and forth can stop. AMD finally admitted a firmware defect and is working on a patch.
Yep! See above. I added in the excerpt to my post. :)

I wonder what those who thought otherwise will say now? Do you think we will hear from anyone after this?
Posted on Reply
#212
Nordic
These amazing cpus are only going to be more amazing once this is fixed.

With an average clockspeed of 4.2ghz my 3900x is performing in the 80th percentile on userbenchmark. Userbenchmark isn't the greatest benchmark, but sometimes aggregate benchmark data can be useful.

I intend to install a full custom water cooling loop later this year, and I am excited to see how high I can get my average clockspeed. Maybe even with this fix amd has planned, I may even exceed the maximum advertised boost clock like that one amd video said may be possible.
Posted on Reply
#213
lexluthermiester
R-T-B, post: 4109084, member: 41983"
Actually, I'd bet the AGESA package may be doing it, but I could be wrong.
That is entirely possible.
R-T-B, post: 4109084, member: 41983"
Either way, it does sound like a software solution is possible. My issue is it should work from day 1... this is a basic, advertised spec.
While I agree, I still have to side with the idea that mobo makers aren't getting it right and need to work it out.
Posted on Reply
#214
EarthDog
lexluthermiester, post: 4109601, member: 134537"
While I agree, I still have to side with the idea that mobo makers aren't getting it right and need to work it out.
https://www.techpowerup.com/forums/threads/der8auer-only-small-percentage-of-3rd-gen-ryzen-cpus-hit-their-advertised-speeds.258840/post-4109568
https://www.techpowerup.com/forums/threads/der8auer-only-small-percentage-of-3rd-gen-ryzen-cpus-hit-their-advertised-speeds.258840/post-4109577

The AIBs have nothing to do with it. Please see the links to the previous posts just above yours.
Posted on Reply
#215
lexluthermiester
notb, post: 4109246, member: 165619"
So why doesn't AMD just help them? They've launched these CPUs with some specification, so they must have been able to build a reference system that worked as on the box.
AMD has given them the specs, who knows why they're not following them. Or maybe they just don't understand everything. It has happened.
Posted on Reply
#216
GeorgeMan
TheLostSwede, post: 4108679, member: 3382"
Care to share the rest of your hardware, as well as UEFI version? Without that, it's hard to give any suggestions.
Of course. It's the Ryzen 3600, MSI B450 Tomahawk with the AGESA 1.0.0.3AB bios. Rest of system is 2x16GB TridentZ 3200CL15 with Samsung B-die ICs and an EVGA 1080Ti, custom watercooled too. You can see some pictures of the real system here. It doesn't matter if I choose Cool & Quiet, PBO enabled or disabled, not even if I set custom limits on the EDC etc. Everything (including full auto settings) results into 4050-4100MHz on single core workloads, max. VRM temperatures are non-issue too, they max out at ~50°C.
Posted on Reply
#217
lexluthermiester
EarthDog, post: 4109607, member: 79836"
The AIBs have nothing to do with it. Please see the links to the previous posts just above yours.
TLDR, and to be fair there is alot of conflicting info available. I'm falling back on what I have observed and been able to make work. I have observed boards applying too much voltage and the CPU's running hot as a result. Lowered the voltage, problem solved. As voltage is applied by settings in the UEFI of the boards in question, the boards made by AIB's are directly responsible. Therefore it is logical to conclude that the AIB's are not getting things right.
Posted on Reply
#218
EarthDog
lexluthermiester, post: 4109623, member: 134537"
TLDR, and to be fair there is alot of conflicting info available. I'm falling back on what I have observed and been able to make work. I have observed boards applying too much voltage and the CPU's running hot as a result. Lowered the voltage, problem solved. As voltage is applied by settings in the UEFI of the boards in question, the boards made by AIB's are directly responsible. Therefore it is logical to conclude that the AIB's are not getting things right.
You really should read those before making more posts. :)

It isnt a TLDR. Those are links to a single post with a paragraph from AMD stating they are fixing the problem through firmware. They dont mention rogue AIB UEFIs or too much voltage or whatever else you've mentioned.

This is not an AIB problem.
Posted on Reply
#219
Chrispy_
I'm not convinced there is a real firmware fix coming. Some minor tweaks perhaps but I suspect the announcement is just damage control to shut up the vocal minority who are making a big deal about this.

There's enough of a spread in the Der8auer survey results to show a clear bell-curve of results implying that this isn't a firmware limitation but simply the spread of results from the silicon lottery. The peak of the bell curve is typically 25-50MHz lower than AMD's figures and if the survey data is realistic then AMD either miscalculated slightly or rounded up the figures to the nearest 0.1GHz.

It's still comical that this topic has even come up, firstly because Intel's CPUs have arbitrary time-limits to their boost, after which they slow down again far more than Zen2 chips do, and secondly because the number of situations where only one core is active in a modern machine is zero. The only people who care about this "peak single-core boost frequency" aren't people who are actually using the chips to do stuff. The minute you give any multi-core CPU a real-world workload, the OS scheduler is going to use all available cores to run background tasks, meaning that 'single core' is never achieved.

Hell, the monitoring software uses a core to monitor the single-threaded synthetic load, thus using a second core. It's so dumb that the only people left arguing it seriously are just in it for the arguing, not actually giving a damn about the topic at all ;)
Posted on Reply
#220
lexluthermiester
TheLostSwede, post: 4109277, member: 3382"
A little snipped of information I just got. It would seem AMD doesn't have a solution to the problem yet, at least not one they've communicated to the board makers, so it might be some time before this is resolved, if it can be 100% resolved that is.
Your conclusions are incorrect. This is easily solved by lowering voltages.

EarthDog, post: 4109624, member: 79836"
You really should read those before making more posts. :)
Why? I have and am solving the problems. If AIB's would lower the default voltages, the problem would be solved. AMD does not need to do anything other than direct this action..
Posted on Reply
#221
TheLostSwede
GeorgeMan, post: 4109622, member: 164481"
Of course. It's the Ryzen 3600, MSI B450 Tomahawk with the AGESA 1.0.0.3AB bios. Rest of system is 2x16GB TridentZ 3200CL15 with Samsung B-die ICs and an EVGA 1080Ti, custom watercooled too. You can see some pictures of the real system here. It doesn't matter if I choose Cool & Quiet, PBO enabled or disabled, not even if I set custom limits on the EDC etc. Everything (including full auto settings) results into 4050-4100MHz on single core workloads, max. VRM temperatures are non-issue too, they max out at ~50°C.
Old AGESA could be part of the problem, nothing much you can do about it until MSI releases an update though. I didn't hit the right speeds until the second beta UEFI on AGESA 1.0.0.3ABB from Gigabyte.
Posted on Reply
#222
EarthDog
lexluthermiester, post: 4109630, member: 134537"
Your conclusions are incorrect. This is easily solved by lowering voltages.


Why? I have and am solving the problems. If AIB's would lower the default voltages, the problem would be solved. AMD does not need to do anything other than direct this action..
Denial is not just a river in Africa (or a city in Ohio according to those opioid commercials, lol!).

It was straight from AMD. If the AIBs were to blame, you're damn right AMD would have said so. They didn't.
Posted on Reply
#223
TheLostSwede
lexluthermiester, post: 4109616, member: 134537"
AMD has given them the specs, who knows why they're not following them. Or maybe they just don't understand everything. It has happened.
Reading comprehension once again. Holy...
If you actually read AMD's email, quoted above, it says they have a firmware bug, no spec in the world would help the board makers work around that, as they can't edit AMD's firmware.

lexluthermiester, post: 4109630, member: 134537"
Your conclusions are incorrect. This is easily solved by lowering voltages.


Why? I have and am solving the problems. If AIB's would lower the default voltages, the problem would be solved. AMD does not need to do anything other than direct this action..
Right, because you and only you, have a solution to all the problems so many of us have had...
How simple, amazing...
I wish I would've tried that three months ago...
Oh right, if I drop my CPU Voltage, my system won't boot...
Posted on Reply
#224
Vlada011
:) I would like to see how Intel i7-6950X compete with new R9-3900X.
Because we talk about 4 years old CPU with lower frequency it's logic to OC both to the maximum and then to compare them.
That mean i7-6950X 4.4GHz boost on all cores, 3.8-4.0GHz Cache frequency vs R9-3900X on how much is boost...
no one know that for sure, enthusiasts community still examine is it boost as AMD advertised.
Posted on Reply
#225
TheLostSwede
Chrispy_, post: 4109627, member: 185623"
I'm not convinced there is a real firmware fix coming. Some minor tweaks perhaps but I suspect the announcement is just damage control to shut up the vocal minority who are making a big deal about this.

There's enough of a spread in the Der8auer survey results to show a clear bell-curve of results implying that this isn't a firmware limitation but simply the spread of results from the silicon lottery. The peak of the bell curve is typically 25-50MHz lower than AMD's figures and if the survey data is realistic then AMD either miscalculated slightly or rounded up the figures to the nearest 0.1GHz.

It's still comical that this topic has even come up, firstly because Intel's CPUs have arbitrary time-limits to their boost, after which they slow down again far more than Zen2 chips do, and secondly because the number of situations where only one core is active in a modern machine is zero. The only people who care about this "peak single-core boost frequency" aren't people who are actually using the chips to do stuff. The minute you give any multi-core CPU a real-world workload, the OS scheduler is going to use all available cores to run background tasks, meaning that 'single core' is never achieved.

Hell, the monitoring software uses a core to monitor the single-threaded synthetic load, thus using a second core. It's so dumb that the only people left arguing it seriously are just in it for the arguing, not actually giving a damn about the topic at all ;)
So how do you explain that some of us have already had the problem resolved courtesy of an updated UEFI/AGESA? I was as I've explained time and time again in this thread, a hard upper clock limit of 4,400MHz until recently. Now my CPU boosts to 4,525MHz no problem. But hey, I'm just making that up, right? As it's easier to make crap up, like you...

Oh and it's also on AMD's official Twitter account now.
[MEDIA=twitter]1168901636162539536[/MEDIA]
Posted on Reply
Add your own comment