Every Ryzen 5000 CPU owner knows, that PBO is not reliable in terms of stability.
For those unaware of the unfixable issue, the combination PBO + CPB results in an intermittent system crash while it is in low CPU usage state, so, ratter you like/want it or not, the only two viable ways to "solve" it, are: Disable PBO + CPB combination, or, manually set the clock speeds.
Well, none of these scenarios are acceptable because AMD sell's out that PBO is a must, and in fact, is the only way to achieve the labelled boost clocks.
The point here is, if you ask for RMA, you will probably receive a new CPU which will show the same behavior, because this is not an isolated issue, this problem happens to be part of the Zen 3 architecture. So, while AMD allows easy RMA for those facing this situation, this "effort" is only to mask the real problem.
EVERY RYZEN 5000 CPU has the PBO + CPB problem and AMD CAN't solve it through firmware upgrades. Their main feature doesn't work as designed and there is nothing to do about it, AMD knows, but wont say a word about it.
Why am I saying so?
Of course, the current production process may be improved and yes, that could be reasonable for a new revision launch. But, considering the effort of the new production and logistics for this revision, we can say precisely assume, that we have more in game, then what is being said. AMD will deliver a PBO error free revision of the Ryzen 5000 cpus, so the market wont be flooded with only the problematic revision.
You may be facing intermittent system crashes while in low cpu usage situations, and yes, if you are a Ryzen 5000 owner, PBO is probably the cause.
These are Windows events related to the problem:
My intention here is to alert everyone that AMD is not being honestly regarding to this issue. I was able to test 3 different cpus in 3 different systems and all of them crashed when PBO + CPB were in use. A 5800x, 5900x and a 5950x, 2 Asus motherboards and an Aorus one.
Every owner has the right to use the features they paid for, and that is not the case with Ryzen 5000 chips. Simple disabling PBO + CPB is not a viable solution.
We should ask for an official AMD statement about this! Or, Justice should be called for an revision change, free of cost replacement of every defective Ryzen chip for those with the issue (including me) Sorry for some English mistakes.
Interesting thread so have subscribed.
I experienced exactly the same issues you describe and am currently in the process of a RMA - interesting to know that this will not fix things but will see what they send back.
In your case, it may be something beyond memory. You need to investigate. Blue screen is something related to the driver or device. In some specific cases an application that may have replaced an important dll for the system.
Give Malwarebytes a visit beforehand to check if it is clean.
Hi!, in my case, my cpu is at this moment at the DHL facility waiting to be picked up. It's been 6 days there since it arrived on sept 3rd.
All the documents were filled out and sent to me by the RMA company I assume is contracted by AMD, in my case is SIR Company in Florida, USA.
I'm also assuming that the CPU will be picked up sometime soon, and then the RMA will start, hopefully, confirming my issue and sending a brand new processor. Everything was done with no cost whatsoever up to this point for me.
In my case, RMA was guided by SIR Company as well. However, AMD decided not to give another sample, instead, I was refunded and bought a Z590 Platform.
When requesting an RMA for a WHEA issue, what if it's highly intermittent and very difficult to replicate? Mine has the exact issues described but replicating it can takes days as it only happens at very specific/idle moments. I am currently requesting an RMA with the retailer I purchased from (scan) and I am worried they will simply plug it in, run a prime test or something similar and call it a day. I have explained to them it is intermittent etc. but I am still worried they will just reject it if they cannot reproduce the issue.
I would imagine any testing undertaken by Scan would be different to that of AMDs.
You should have raised a RMA dirrctly with AMD to avoid Scan's process, as AMD's testing takes place in amsterdam.
Sounds like your return is going to be tested by Scan to ascertain/replicate your fault
Well, after explaining to Scan that it is very difficult to replicate the issue and is intermittent in nature etc they have still agreed to the RMA. I am hoping someone doesn't just plug it in, boot up cinebench and go "yeah, looks like it's working to me". Hopefully I will have good news to share in the coming days...
I am curious - for those that submitted their CPUs for RMA to AMD because you were hitting WHEA under PBO conditions, did you tell AMD that and did they accept the RMA? As far as I understand, PBO voids the warranty? So, if the only situation you encounter a WHEA is under a warranty-voiding situation, why would they accept the RMA? Unless you simply didn't tell them? But surely if it works with PBO disabled, they would test and verify that and then reject the RMA anyway?
@NathObeaN - Whea happens even with the PB (non overdrive) which is a supported feature. Also, during chipset installation, AMD recommends you to use PBO for improved performance, so, they don't have much to complain since they rely on this feature to have the boost function.
In my case, I've mentioned some threads about WHEA-18 with them and, they decided to give my money back. Thanks to that, I have a FULLY STABLE INTEL + NVIDIA System without a SINGLE error.
"But mommy, it requires me going to BIOS, offing useless auto-clock features and setting the desired multiplier and voltages manually, it's COMPLICATED for my smol brain, I want my processor to overclock itself, waaa!.."
@EFermi not sure what you’re referring to? I’ve been overclocking computers for the past 20 years. I don’t find a single thing about Ryzen ‘too complicated’ to overclock. I (as others) prefer the idea of PBO + curve + scalar etc because it’s arguably a more complicated way to overclock then simply stickikg a static clock + voltage and calling it a day, but it’s far more efficient to varying work loads.
Ok so I have an update - after going through an RMA for a new 5950X all my problems have gone. No more WHEA. The new CPU is stable and boosts higher. Seems rock solid. Glad I went through the RMA process
After 17 weeks waiting for my system to have parts available i am getting the same errors you are on a brand new ryzen 5900x on a Rog Crosshair 8 Hero. I spent a day on the phone with the sysstem builder's support testing and stressing everything.
but ofc when it's being tested it's fine, after all the stress tests driver shecks, system intergity check...i had a chrome window open, singlke tab, wlaked away for a coffe and come back to blank desktop as it BSODs almost every f((*&ing time it's NOT doing anything taxing.
Can you guys suffering from WHEA errors try something for me, based on a hunch? In the advanced power settings, could you change the minimum processor power to 10%, or something higher than 5%?
Settings > System > Power & Sleep > Additional Power Settings > Change Plan Settings > Change Advanced Power Settings > Minimum Processor State. Set this to 10%
Another today. Tired of this. First and last AMD build. Crap Company. Starting RMA process, or full money payback for the whole combo. Switching to Intel.
Contacted support from the system builder, very helpful person. Described exsactly what i have seen here and experienced personally and got their help to turn off CPB and PBO dropping CPU sped to stock with no boost. No other changes and machine hasn't WHEA'd in 10 days of use both heavy and idle.
He sent me a system builder BIOS from ASUS that's tweaked and not beta (newer than 3204 version currently on machine) but i haven't tried it yet.
I'll post if i get that installed and test out turning boosts back on
very NOT happy with AMD right now
If CPB and PBO need to be off for it to work, it isn't working as advertised, that's for sure. I returned my 5900x, I got sick of all of this. The only way mine worked was to disable CPB and PBO.
@3DJF - Can you share your Bios? What motherboard is this BIOS for?
Very shame, AMD. ASUS is solving this issue their own way.
Well, without CPB/PBO the only thing left to reach higher clocks is manual overclocking, and this is not what AMD sell. Their boost must work, simple.
I'm also in RMA process, submitted my 5950x today. Let's see what I get back.
@digriz - Yeah, this workaround is known, and not viable. We paid for this boost, so, it must work.
AMD is probably hiding because this issue affects all chips, the point it that most people doesn't know what is affecting them.
I agree--I would prefer all the necessary boosting and things that I paid for and expect to just work. I don't want to tinker around with overclocking because I don't know how and I don't want to damage anything!
Ugh, glad you found a solution, but not happy that it involves turning off the autoboost feature. The whole point of it was to give me maximum safe performance without needing me to tinker with it (since I don't want to ruin anything).
I don't even know if an RMA'd CPU would have this issue fixed, or if it's going to plague all chips (in my case, 3700x).
Update: RMAed my 5950x in May this year and after discussing with AMD support about WHEA, they decided not to give me another CPU, instead, they gave refunded me.
With the money, I've got a Z590 + 10900k and I'm now free of WHEA and USB issues.
I sincerely recommend you all, to do the same. AMD will NEVER take over the fact they flooded the market with a defective architecture and due to the poor quality of the reviewers Worldwide (that only compare performance but not stability and/or usability) we were not aware about this flaw, before getting a Ryzen 5000 series CPU.
Wish you the best. Intel for life!
Nope, in electronic products there is high RMA rate for some items, e.g. motherboards.
You are implicating that because of you and a couple of dudes have had crashes that ZEN architecture is defective? Complete nonsense.
Intel is a waste of sand currently, nothing more. We will see what happens at 3nm.
my latest bios for aorus pro ac drop the turbo boost from 4850 to 4800.... like ly cause by the motherboard push too high for amd.... some chip is power and some chip is weaker... and this whea happen random........
turbo boost should be 4700 and board still boost it higher.........
Hello everyone, I go to buy back in April go to buy a 5900X but due to the GPU shortage I finally got to complete it 2 weeks ago. Since then I have experienced those WHEA reports with BSODs as a result. This is effing frustrating, I did not know about this problem at all, to say that I'm outraged is not enough. Now, after reading this topic I have come to the realization that even if I do an RMA it could still happen. But my big problem is that I bought my CPU through AMD.COM and sent it to my country, Perú, South America. I didn't understand why it was happening cause then I ran PRIME95, CINEBENCH, FURMARK no errors were shown, but a single Chrome tab would crash the system.
Now I have to do an RMA without knowing that it will go through and get accepted cause my CPU is in Perú and not in the US.
But it is true, if you overclock with RYZEN MASTER, it would be OK, but boost would stay on all the time and that isn't what I want because of the temperatures, even though I have a 360 AIO.
Is there a workaround?, this is my last hope.
Thank you in advance for your kind help.
PS.- my system is a 5900X - MSI RTX 3080 Ventus 3X - Corsair Vengeanse RGB 8GB x 4 - Corsair iCue 5000D Airflow - 1 TB SSD x2
you have aio... just set manual voltage and manual cpu speed........ you can run even faster than the boost stuff....
i have no aio and tweak for curve for month and still can get errror on the gold selected core... maybe need positive the curve
@ebisettia - If you're not considering to RMA your parts, specially the CPU, the only stable recommendation is the fixed clock usage. I know how repulsive it feels nowadays, however, that's the way to go, if you face WHEA while using PB/PBO.
In my case, I've used my 5900x with fixed clock for 4 months, and, despite the USB problems, I was able to use it daily, without concerns.
But, if PBO is a must for you, the RMA can be tried, however, the chance to get another defective sample is high.
About your RMA concerns, as long you have the proper Serial Number, AMD RMA is available worldwide.
@Electric_Squall I'm gonna try RMA first with a replacement part, but if I keep getting WHEA errors, I RMA again but with a refund. It is really a sad thing to find out that such an awesome product can have this kind of issues.
I've been watching videos about overclocking, playing with CO, maybe I can get something more optimized than the auto value and stop those WHEA errors, who knows. Right now my PC is at the seller's service center, but I know they will tell me that it is the processor that is faulty and that part I didn't buy it from them.
Thank you for your kind help.
Have a great one.
@ebisettia - Hello buddy, how are you?
I also consider that RMA is the best approach for this situation. Unfortunately, it is very unlikely that someone that pays the price to have a 5950x will ever accept unstable features, which is the case of PB/PBO for these chips.
Recently, I've read that some people that said CO with positive offset to all cores, can "workaround" the issue. Those who said so, pointed out that values of +3 applied prevented the system from "WHEAS" Well, I don't how reliable is this information because WHEA may happen quite often, or only after some considerable time (+3 weeks) so it is hard to ensure CO with positive offset can really fix the problem.
But this is it, I think that PB/PBO should work out of the box, period.
Wish you the best my friend.
You can consider this information (setting CO into the + as a workaround) reliabile. I've been living with CO +10 on all cores for several months already without any noticed crashes or reboots (there were some crashes but they were related to the video driver).
I thought +8 was stable for me.
Just find the CO value which is stable for you for a week and add another +2.