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

#51
ZoneDymo
bug, post: 4108683, member: 157434"
No, that is not what boost has meant until Zen2. If that were the case, you'd see manufacturers using "1THz boost" on their boxes, because, well, it's "up to" and one of their engineers has seen one CPU reaching those speeds when no one else was looking.

And the debate is not useless, like the debate about GTX 970's VRAM wasn't useless at the time: it doesn't mean users get screwed, but it sends a clear message this is a practice we could do without.
Now you are being just silly, obviously in this case the "up to" speed can and has been reached by multiple CPU's of the type.
Nothing to do with that honestly kinda childish scenario you are conjuring up.

The fact is that is just down to cooling, if you have enough cooling you can reach the needed speeds (even if some manual tweaking might be involved).
Its what Steve from GN has been saying for a while now, OCing is seemingly dying, the modern processor will push itself as far as your cooling (and stable voltages) will allow.

If Nvidia put on the box of the GTX970 "up to 4gb of Vram" they might have been in the clear on that one, because im pretty sure everyone would look at that and think hmmm bit of a gamble and then make the choice to buy or not.
Posted on Reply
#52
bug
Vya Domus, post: 4108734, member: 169281"
So this works one way only huh ? You do realize this is literally just placebo, right ? No matter what you do some people will get less, some more, no matter which way you set the balance. This is bound to cause complaints no matter what. Arguing about this is simply a waste of time, the only thing that would make this somewhat fair is that everyone stops stating "boost clocks" on their products and only claim a certain base clock.
Yes, this only works one way.
It's like when you buy food. When it says it's safe to eat it within 3 months, it can happen you can eat it after 4. But in no way will it go bad in less than 3 months if stored properly.
Posted on Reply
#54
bug
ZoneDymo, post: 4108747, member: 66089"
The fact is that is just down to cooling, if you have enough cooling you can reach the needed speeds (even if some manual tweaking might be involved).
No, you can't. Several users here have attested to that.
Posted on Reply
#55
ZoneDymo
EarthDog, post: 4108741, member: 79836"
Funny because big bad NVIDIA does it RIGHT listing a GUARANTEED boost clock (minimum) and not listing a maximum!!!!!

Look at gpuz... the boost clock. And 100% of these will achieve that clock and typically 100-200MHz more when using boost (assuming no power virus running against it).

Dont act dense... you know Intel actually hits these boost clocks on all cpus and that is a piss poor argument, right?
ermmm no? the stock intel cooler will thermal throttle modern day cpu's hard sooo yeah.
Posted on Reply
#56
Darmok N Jalad
Just as an aside, I notice that with HWiNFO, there is a lag in readings when I apply a load. I’m not sure what the polling interval is, but it’s clear that it doesn’t capture the CPU state for every clock cycle. This is still why I would like to see what AMD says. I think there is silicon lottery, but there may also be very brief (millisecond) boost instances. I know mobile SOCs employ this strategy considerably, in order to stay within power and cooling budgets.


Hossein Almet, post: 4108749, member: 168115"
Does the 9900K hit 5GHz with a $30 cooler?
According to Anandtech, yes, in single core. You still have to use at least a 95W cooler, and doing so will result in 15-20% less multithreaded performance over a more powerful cooler.
Posted on Reply
#58
EarthDog
Hossein Almet, post: 4108749, member: 168115"
Does the 9900K hit 5GHz with a $30 cooler?
lol, yes. Hyper 212 evo?
ZoneDymo, post: 4108754, member: 66089"
ermmm no? the stock intel cooler will thermal throttle modern day cpu's hard sooo yeah.
not 2 cores and 5ghz.. the advertised boost clock amd number of cores. Perhaps it throttles all core/threads on the stock cooler... but boost clock is 2 cores.

But you've also missed the point...even with the right mobo and bios, as well as a proper cooler, MOST STILL DONT TICK MAX BOOST. These arguments are ridiculous and sourced from misinformation.
Posted on Reply
#59
Jism
bug, post: 4108751, member: 157434"
No, you can't. Several uses here have attested to that.
They need to do some tweaking first. People are running into FIT issues. Really. Everybody seems to blate over the net, nobody that took the effort to first of all, set HWInfo on a 200ms interval. Yes the default interval of HWInfo is like 500ms or even 1000ms. When you see your CPU core or cores run into the 60 degrees mark, shit starts to throttle down. If a single core has not reached the max boost speed, really what would happen if you slightly start undervolt your CPU? Has anyone tested that of you?

FIT is preventing the CPU from boosting to max advertised. Im sorry but is the majority of testers stupid?

And second, a stock cooler MIGHT hit boost for a small short of time, but it will dump when theres a serious MT workload applied.
Posted on Reply
#60
ZoneDymo
EarthDog, post: 4108758, member: 79836"
lol, yes. Hyper 212 evo?
not 2 cores and 5ghz.. the advertised boost clock amd number of cores. Perhaps it throttles all core/threads on the stock cooler... but boost clock is 2 cores.

But you've also missed the point...even with the right mobo and bios, as well as a proper cooler, MOST STILL DONT TICK MAX BOOST. These arguments are ridiculous and sourced from misinformation.
and where do you get your information then?
Posted on Reply
#61
cadaveca
My name is Dave
EarthDog, post: 4108758, member: 79836"
lol, yes. Hyper 212 evo?
not 2 cores and 5ghz.. the advertised boost clock amd number of cores. Perhaps it throttles all core/threads on the stock cooler... but boost clock is 2 cores.

But you've also missed the point...even with the right mobo and bios, as well as a proper cooler, MOST STILL DONT TICK MAX BOOST. These arguments are ridiculous and sourced from misinformation.
I'm sorry but I KNOW yo are kind of missing the point here too. Like you are right, and you and I usually agree on most things, so when I see what's going on here, I see that what the cause of the problem is... and it's the BIOS. One update, and everything is working as it should.


So here we are with new AMD CPUUs that need an AGESA update before they are working as they should? :eek: NO WAY!


:kookoo:

ZoneDymo, post: 4108765, member: 66089"
and where do you get your information then?
He does reviews and has access to information the general user does not. One of the few guys doing reviews that truly isn't biased, IMHO, so if he's taking the time to type something to here, there's something that should be looked at, IMHO.
Posted on Reply
#62
Metroid
Well as we can see 26% of 3800x hit the adv speed boost, 6% of 3900x hit the adv speed, that means in theory that all the best silicon will be going to 3950x. So my point is that, if you want the best then wait for 3950x, avoid 3900x. I think 3950x will be the only one to have more than 50% to hit adv speed boost 4.7ghz. So what to hit the adv speed boost? then stick to 3600 or 3950x, yes, is that simple.
Posted on Reply
#63
Vya Domus
Darmok N Jalad, post: 4108756, member: 170588"
Just as an aside, I notice that with HWiNFO, there is a lag in readings when I apply a load. I’m not sure what the polling interval is, but it’s clear that it doesn’t capture the CPU state for every clock cycle. This is still why I would like to see what AMD says. I think there is silicon lottery, but there may also be very brief (millisecond) boost instances. I know mobile SOCs employ this strategy considerably, in order to stay within power and cooling budgets.
I can see the minimum polling rate is 100ms, that's pretty high and CPUs can boost for very short intervals.
Posted on Reply
#64
EarthDog
cadaveca, post: 4108766, member: 25138"
I'm sorry but I KNOW yo are kind of missing the point here too. Like you are right, and you and I usually agree on most things, so when I see what's going on here, I see that what the cause of the problem is... and it's the BIOS. One update, and everything is working as it should.


So here we are with new AMD CPUUs that need an AGESA update before they are working as they should? :eek: NO WAY!


:kookoo:
Hi Dave! I hope you are well! :)

So, as I've said earlier... I've used at least 6 boards (more now I think, lol) with multiple bios and agesa's. Nada. They've been out for almost 3 months.. maybe this will all end with another update...sure. But it hasnt yet. ;)
Posted on Reply
#65
Jism
Vya Domus, post: 4108768, member: 169281"
I can see the minimum pooling rate is 100ms, that's pretty high and CPUs can boost for very short intervals.
Its 100 or 200 ms indeed. But a Ryzen CPU can update or refresh every 25ms. Thats 4 to 8 times the maximum polling rate Hwinfo can.
Posted on Reply
#66
TheLostSwede
EarthDog, post: 4108716, member: 79836"
Cause you bought on a BS premise? I've got a 3700x that doesnt hit boost clocks...isnt that midrange being 3 products down from the top and two from the bottom?
Which is how I felt as well, until it was resolved with a new UEFI. In other words, there's hope, but the silence from AMD on the issue is aggravating and they could do a lot better in terms of communicating with their customers on this issue.
Posted on Reply
#67
EarthDog
ZoneDymo, post: 4108765, member: 66089"
and where do you get your information then?
Seemingly in a better place than you... :p
TheLostSwede, post: 4108775, member: 3382"
Which is how I felt as well, until it was resolved with a new UEFI. In other words, there's hope, but the silence from AMD on the issue is aggravating and they could do a lot better in terms of communicating with their customers on this issue.
Sure there is hope! That just wasnt part of the conversation until now.

People are trying to justify it and come from a misinformed place thinking that if AMD's parameters are met it works. It does not. Not for many... and dare we say a majority.

Maybe 1.0.0.4 is my magic sauce... we'll find out shortly.
Posted on Reply
#68
cadaveca
My name is Dave
EarthDog, post: 4108771, member: 79836"
Hi Dave! I hope you are well! :)

So, as I've said earlier... I've used at least 6 boards (more now I think, lol) with multiple bios and agesa's. Nada. They've been out for almost 3 months.. maybe this will all end with another update...sure. But it hasnt yet. ;)
I'm doing excellent. I wish I could say more, but... I hope you are doing well too! I'm still here, of course.

I think AMD needs 4-6 months before they can suss out all the little niggles I see. I mean... do we even have odd CAS numbers working yet? ;)

So like, this is a real problem, there can be no doubt. I don't see it as that big of a problem (as is usual for me and this stuff), but it is something that consumers should be made aware of in a better way. I'm not sure that this is the correct approach, as we have to filter out the mail carrier's bias as always.

TheLostSwede, post: 4108775, member: 3382"
Which is how I felt as well, until it was resolved with a new UEFI. In other words, there's hope, but the silence from AMD on the issue is aggravating and they could do a lot better in terms of communicating with their customers on this issue.
Lisa Su said that'd get better, and that's the one area that I think AMD could really improve upon. but there's too many large egos in the media that get in the way of that ever being an easy task. AMD jsut doesn't use what most consider as traditional avenues to reach their end users because of this, for example. How do you fix that? You just can't.
Posted on Reply
#69
Jism
EarthDog, post: 4108776, member: 79836"
Seemingly in a better place than you... :p
Sure there is hope! That just wasnt part of the conversation until now.

People are trying to justify it and come from a misinformed place thinking that if AMD's parameters are met it works. It does not. Not for many... and dare we say a majority.

Maybe 1.0.0.4 is my magic sauce... we'll find out shortly.
AMD should just put that extended advertised boost clock under a warranty notice, so that if you CPU starts degrading over time because you and a dozen other users demand a up to frequency, their complete liability is just not theirs anymore. I just commented on Derbauers vid, i know its not passing their QOS moderation lol, i dont care.
Derbauer, you are plain stupid. If youve followed any information from the Stilth, you know that these CPUs are equipped with a FIS that is hardware technically preventing the CPU cores from consuming too much current. The FIS also is a silicon health feature that prevent the boost state from going into oblivion when the CPU starts to hit a unsafe or unstable condition. The SKU advertises clearly with Base clock and Boost clocks up to. Up to, which doesnt mean you will be getting the guaranteed boost clocks. There is variance in motherboards and variance into used biosses as well. If i take my 2700x which goes into boost state on IBT for barely a minute on 4.1GHz, it seems to start holding when i increase my cooling and apply a undervolt for up to 5 minutes on 4.2Ghz all core. I woudt have made this without better cooling and applying a undervolt. You apply your survey based on stock settings and a majority of users going stock. There was no tweaking involved either. I know this comment isnt going to pass your moderation, but really your creating FUD and just putting garbage on the net the way the boost state works.
Im kind of done by now commeting and trying to give the masses some understanding on how the XFR works on AMD cpu´s.
Posted on Reply
#70
Vya Domus
Jism, post: 4108772, member: 91255"
Its 100 or 200 ms indeed. But a Ryzen CPU can update or refresh every 25ms. Thats 4 to 8 times the maximum polling rate Hwinfo can.
And the default value on mine was a staggering 2000ms, good luck keeping a CPU under the right conditions for single core boost clock to work for two seconds.

I've done benchmarks many times playing with these things and sometimes it would never pick up the 3.9 Ghz max clock my 1700X is supposed to have. I knew it actually runs at those speeds because the results are higher versus when turbo clocks are turned off but sometimes I would never witness the clocks going up.
Posted on Reply
#71
EarthDog
cadaveca, post: 4108777, member: 25138"
I'm doing excellent. I wish I could say more, but... I hope you are doing well too! I'm still here, of course.

I think AMD needs 4-6 months before they can suss out all the little niggles I see. I mean... do we even have odd CAS numbers working yet? ;)

So like, this is a real problem, there can be no doubt. I don't see it as that big of a problem (as is usual for me and this stuff), but it is something that consumers should be made aware of in a better way. I'm not sure that this is the correct approach, as we have to filter out the mail carrier's bias as always.



Lisa Su said that'd get better, and that's the one area that I think AMD could really improve upon. but there's too many large egos in the media that get in the way of that ever being an easy task. AMD jsut doesn't use what most consider as traditional avenues to reach their end users.
Most of us will be waiting with bated breath. :)
Jism, post: 4108779, member: 91255"
AMD should just put that extended advertised boost clock under a warranty notice, so that if you CPU starts degrading over time because you and a dozen other users demand a up to frequency, their complete liability is just not theirs anymore. I just commented on Derbauers vid, i know its not passing their QOS moderation lol, i dont care.
But, we haven't demanded anything more than they offered...If the CPU degrades up there, then why the hell would they advertise it?
Posted on Reply
#72
Jism
Vya Domus, post: 4108781, member: 169281"
And the default value on mine was a staggering 2000ms, good luck keeping a CPU under the right conditions for single core boost clock to work for two seconds.
Really, its just FUD. in real life workloads the things that i do with my PC, its just boosting as it should, while being silent, and having a balanced power plan going on.

Just use your computer the way you intended, instead of going by the masses of derbauer and his follower. FUD creating in exchange for visits and clicks. There are people that deserve way more credit for their work then Derbauer.
Posted on Reply
#73
TheLostSwede
Vya Domus, post: 4108745, member: 169281"
No we aren't, we are speculating that may be the case, based on some pools.
You seem to like water a lot...
The word you're looking for is polls ;)
Posted on Reply
#74
Vya Domus
TheLostSwede, post: 4108786, member: 3382"
You seem to like water a lot...
Not much into water cooling though.
Posted on Reply
#75
TheLostSwede
ZoneDymo, post: 4108747, member: 66089"
Now you are being just silly, obviously in this case the "up to" speed can and has been reached by multiple CPU's of the type.
Nothing to do with that honestly kinda childish scenario you are conjuring up.

The fact is that is just down to cooling, if you have enough cooling you can reach the needed speeds (even if some manual tweaking might be involved).
Its what Steve from GN has been saying for a while now, OCing is seemingly dying, the modern processor will push itself as far as your cooling (and stable voltages) will allow.

If Nvidia put on the box of the GTX970 "up to 4gb of Vram" they might have been in the clear on that one, because im pretty sure everyone would look at that and think hmmm bit of a gamble and then make the choice to buy or not.
But that's EXACTLY the point, it's NOT just down to cooling. My CPU couldn't go 1MHz past 4,400.25MHz at one point, along comes a new UEFI and now it boosts to 4,525.25MHz on multiple cores. So in other words, people not hitting advertised boost, is NOT related to cooling alone. Don't try to simplify this issue, as it's really a lot more complex than that.
Posted on Reply
Add your own comment