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:
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.
If you would like to see the more in depth testing, here is the English version of the video:
253 Comments on Der8auer: Only Small Percentage of 3rd Gen Ryzen CPUs Hit Their Advertised Speeds
HOT SUMMER.
www.techpowerup.com/forums/threads/ryzen-3000-boost-issue-whats-your-take.258696/page-4#post-4108389
It's obvious a lot of people who've submitted data either don't know how to find out the peak boost speed of their CPU, or they have some other issue preventing their CPU's from boosting.
Obviously a large amount of users are having the same kind of issues that most of us have had/are having as well, but there's also a lot of suspicious data. On the other hand, it also proves that a lot of people are stuck at ~100MHz below advertised max boost, for some strange reason.
In all fairness, der8auer filtered out a lot of the crap results.
Not sure I fully agree with his conclusion, but yeah, AMD really needs to go out there and clarify things, as there's too much speculation and too little factual information with regards to what's going on.
This is why I'm saying this is about as accurate data as that from an election poll, as there are too many variables and not enough information provided.
Another things that's missing from this data is, for how long periods of time did people try to reach the boost speeds? 1 minute, 5 minutes, an hour? As it's a "random" occurrence in a way, a longer period of time is needed to actually see if the CPU boosts or not.
On the other hand, a lot of users seem to have the same issue I had for the longest of times, the max CPU core boost is stuck at around 100MHz below AMD's claimed boost. My issue got resolved with a UEFI update, so hopefully more people will be in the same situation and it can be easily rectified.
Then again, I'm surprised a small percentage of users are boosting beyond the max core boost, which implies that AMD's video where they claimed an extra boost with certain hardware configurations, might indeed be possible. I just would like to know what those configurations are...
On the latest ones, RAM is much better but the cores are only seeing about 4500 occasionally and lower clocks most of the time.
Either this is more "observer effect" and/or they need to do some more UEFI tweaking.
It's certainly very strange.
Is there maybe a "fancy new" UEFI/AGESA being readied for 3950X?
My impression is now that AMD has people's hearts, they're feeling safe enough to play fast and loose with specs. Remember the RX 480 TDP debacle?
Intel's Summer of Paying Shills campaign is paying off since nobody on the comments bothered to look at the data and trusted a shill.
What we do know, is that the board makers have different priorities in how they release new UEFI updates as well and plenty of boards still don't have stable AGESA 1.0.0.3ABB UEFI releases. Obviously some of this is on AMD as well, as they've clearly had AGESA related issues too.
It would also be nice if AMD could provide more details on their scheduler and why the fasts or even second fastest cores aren't the ones utilised as the main cores in the CPU. I generally end up loading the slower cores on my CPU, rather than the fast ones. That said, it also seems that my "fastest" core, doesn't boost as high as 2nd, 3rd and 4th fastest cores...
There are too many oddities like this, that AMD needs to come out and explain better.Where did you find the data?
I wouldn't call der8auer a shill though, he has no reason to take money from Intel. However, the way he's applying what he learnt in school on how to extrapolate the data, might not be correct in this case, as anything that's +/-2 Sigma, he simply filters out, which isn't really how you do statistics...
In all fairness, I can't complain any more, as my hardware delivers what AMD claimed on the box.Which is part of the problem as well. It leads to more confusion and more discussions like this.
The box of any AMD cpu states clearly, boost up to Y speeds. Up to. That doesnt mean that any CPU will be capable of doing that.
You have a few parameters that should be taken into consideration:
- Temperatures
- CPU stability (FIT)
- Motherboard current supply
- Type of single-core workload, it all comes down to current drawn by that same single core
I think if we exclude a faulty motherboard, and proper cooling, you would be left with 2 things and those are the FIT CPU stability at that given clockspeed, and the type of workload. FIT kicks in when one core is drawing too much current and puts the boost clocks down. If you would bypass FIT it would fry the internals of the chip (degradation). If the given workload is a lightweight one, i.e not too much current, youd actually see that the boost clock could be held at whatever AMD is advertising with.
Some cores simply are not stable for the max advertised boost speeds, so FIT makes sure it stays within that range to provide a perfectly stable core.
Their system is really clever, the boost guarantees a stable working at a given speed without risk of harming the CPU on long term base. Its the same argument as people would be setting 1.4V manual while the single core speed is reaching up to 1.55V or so because people think it would be hurting their CPU on long term. No dummys putting 1.4V on long term is going to harm you CPU!
I think the masses are not aware of how the boost algorithm works, and how AMD implemented it. Some of you should really go in depth and start monitoring the CPU and read posts like the Stilth who already documented the boost algorithm and behaviour. I spend quite a few hours with my 2700x to know that, a slight undervolt is enough to have it all core on 4.1GHz and in some occasions on 4.2Ghz while on single core having 4.35Ghz is the best i can get with it in combination with a 360mm rad paired with 6 fans. The boost starts to drop once the CPU archieves 60 degrees and beyond. So, for full boost constant try to keep the cpu within 60 degrees. But thats difficult since the core is so small that the heat density is just coming greater and greater as you can see on Intel CPUs too.
Theres not much headroom beyond that what the XFR is already giving you.
Did you see his survey? It's right here - der8auer I need YOUR Help! RYZEN 3000 Boost Analysis Survey
1.) Which CPU do you use ?
2.) Which Motherboard do you use ?
3.) Which AGESA you use ?
1.0.0.2
1.0.0.3 AB
1.0.0.3 ABB
*Notice he didn't include 1.0.0.3 A
4.) Please enter the max boost of your CPU during Cinebench R15
Comment
How can he discard any specialized cooling if he didn't ask for such information in the first place. How can he know if people are or aren't using PBO either. He certainly wasn't asking for any type of proof/validation.
Scientific? More like a questionnaire.
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.
We can debate what "max" means all day long, but we shouldn't have to. They should have been really clear about this. Either in the launch presentation or at the very least on the product page on their own website.
Source:
AMD 3900x