Question Ryzen 5000 series having worrying failure rate

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
20,839
3,174
126
So rumors or statements have been going out about the 5000 series having high failure rates.
At first i threw it off as rumors from PowerGPU due to them pulling the thread.
But now PCMag has also confirmed some troubling issues with the silicon.

I'm not saying AMD is at fault here, i think something might be wrong with TSMC's 7nm.

The problem also does not only lie in the CPU.
I'm also hearing very bad things about the motherboards (B550 and X570).
About USB issues in general.
There are a lot of threads like this popping out on Reddit and other tech forums outside ours:

To me honestly this is sort of disturbing.
I can always write off one thing as just bad building, but when its slowly catching fire and momentum, it has to make you wonder.

Is it memory compatibility or again, is something wrong with TSMC?
Maybe a mouse got in the waffer and thought it was real and took a off it? (sarcasm)
Seriously tho, you guys think this is all just hype or whose at fault here?
TSMC rushing production, or AMD's 7nm arch?
 

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,575
146
So rumors or statements have been going out about the 5000 series having high failure rates.
At first i threw it off as rumors from PowerGPU due to them pulling the thread.
But now PCMag has also confirmed some troubling issues with the silicon.

I'm not saying AMD is at fault here, i think something might be wrong with TSMC's 7nm.

The problem also does not only lie in the CPU.
I'm also hearing very bad things about the motherboards (B550 and X570).
About USB issues in general.
There are a lot of threads like this popping out on Reddit and other tech forums outside ours:

To me honestly this is sort of disturbing.
I can always write off one thing as just bad building, but when its slowly catching fire and momentum, it has to make you wonder.

Is it memory compatibility or again, is something wrong with TSMC?
Maybe a mouse got in the waffer and thought it was real and took a off it? (sarcasm)
Seriously tho, you guys think this is all just hype or whose at fault here?
TSMC rushing production, or AMD's 7nm arch?
RMA rates for Ryzen 5000 are perfectly average for any product generation. No statistics from any system builder/retailer have suggested otherwise, nor have any comments from them with the exception of PowerGPU and one other. The former weren't willing to talk about how they were validating their chips, the latter clearly stated any higher failure rates they were seeing were only when being pushed beyond the stock configuration (i.e. >3200MHz RAM etc) so technically speaking they're not actually faulty chips.

That being said, there is a problem with B550/X570 that AMD have already acknowledged, and you can see this with return rates etc of these motherboards - they're higher than normal.

Still not to the extent where it's a deal-breaker - we're still talking well under 5% of motherboards here. But it is certainly higher than normal.
 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
28,274
19,922
146
Steve from techspot/hardware unboxed said he has spoken with vendors with high volume sales and the failure rates are all very low/within normal margins.

On a personal note: anything that scares people away from buying the 5000 series would be welcomed. Then I could get my hands on one without supporting scalpers. :p
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
It was originally Hassan from WCCFTech who "reported" (read: made up a story) based on a tweet (which has since been retracted). So there is no factual basis for this reporting.

Until I see information from a reputable source or independent investigation of these claims, I will file them away into the "possibly malicious but likely just click bait to drive advertising" category.

That's not to say I'm not irritated at AMD given the persistently poor supply of their otherwise very nice 5000 series CPUs and 6000 series GPUs. So much so that I found an offer I could not refuse for a Intel i5-10400F which is going to be the core (heh) of my next entry-level PC gaming build.
 

DAPUNISHER

Super Moderator CPU Forum Mod and Elite Member
Super Moderator
Aug 22, 2001
28,274
19,922
146
That's not to say I'm not irritated at AMD given the persistently poor supply of their otherwise very nice 5000 series CPUs and 6000 series GPUs.
The worst part is that our crowd/DIYl, is the last in line now. And with the perfect storm of C-19 related supply and demand issues, AMD GPUs, and consoles, all hitting closely together, we get the leftovers. Or at least it feels that way. If I needed a gaming PC or vid card right now, I'd buy prebuilt. Some great deals keep popping up compared to scalped GPU pricing. And I will not support the shameless Ferengi.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
There seems to be a temporary fix for the USB issue, at least from what I read in a thread on reddit. You'd need to disable PCIE Gen. 4 and revert to Gen 3 to 'fix' the problem.
 

dr1337

Senior member
May 25, 2020
311
512
106
I don't see why problems with N7 and x570 would come to the fore now since AMD has been using both since 2019.
I think its something to do with the latest bios revisions. I updated my x570 board just for funsies hoping maybe my 3600 would clock better :p, but for whatever reason the latest bios absolutely decimated USB and was making my sound and keyboard drop out randomly.

Maybe its due to vendor specific implementation? And thats why its only some users on some configurations that have the problem?
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
I think its something to do with the latest bios revisions. I updated my x570 board just for funsies hoping maybe my 3600 would clock better :p, but for whatever reason the latest bios absolutely decimated USB and was making my sound and keyboard drop out randomly.

Maybe its due to vendor specific implementation? And thats why its only some users on some configurations that have the problem?

That's really odd. Is it causing problems with USB ports connected to the chipset, or to the CPU/SoC? I'm using an old UEFI revision from AGESA 1.0.0.3ABBA which dates back to 2019. That might explain why I haven't encountered the issue.
 

moinmoin

Diamond Member
Jun 1, 2017
4,934
7,619
136
If find it odd that this thread is started after the RMA rumors was already debunked and to be even less than for Ryzen 3000 (so it actually improved).

Then combining it with the unrelated valid USB + PCIe 4 issues (that AMD definitely sat on way to long without doing anything) is bad form.
 

Torn Mind

Lifer
Nov 25, 2012
11,567
2,626
136
It seems that it is prudent for others to take one for the team with regards to AMD processors. That does mean waiting quite some time though.
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
Normally I react negatively when I see something that could possibly be "fud," but given the low supply of these CPUs, I heartily approve of anything that might scare away would-be buyers.

I'm still waiting for a 5900X at anything near MSRP.
 

Mopetar

Diamond Member
Jan 31, 2011
7,797
5,899
136
It would be odd for the issue to be due to the TSMC process given that AMD had used it previously. The other post about RMAs actually being lower is online with expectations given a more mature process and additional experience with it.

Every new product will have some issues, but without detailed analysis it's hard to quantify because you don't hear as much from people who have everything working fine or who aren't experiencing any problems.
 

gdansk

Golden Member
Feb 8, 2011
1,981
2,355
136
It's conflating a few issues here. The USB issue isn't the same as CPU/motherboard combinations not booting. I had to update my B550 motherboard to boot the 5900X. Honestly, that alone would probably cause the 3x higher return rates.

As for the USB issue, AMD should take it seriously. I have encountered USB connectivity issues as have a few of my friends. It seems to depend how you use the system but is very really for some people.
 
Last edited:

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
I'm all for taking the USB problem seriously as both disabling PCIe 4.0 and C-states can only be considered a temporary bandage until AMD fixes this irrevocably. However, conflating it with the already debunked rumor of DOA Ryzen chips is not something that should be done on the forum, especially considering the experience and expertise of the OP.

We have a thread on the Motherboards forum on the USB issue, and I'm all for keeping a sticky on the USB issue on the CPU forum as well: it's that important. But let's keep the DOA FUD out of it unless clear evidence of any other problem comes up.
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
20,839
3,174
126
Where do you think the PCI-E lanes come from? Do you think those physical slots work without lanes from the CPU?

Problems with the PCI-E lane tied to the USB can also be a sign of issues with the CPU.
This is why i posted both.

Not to mention both chips are also made from TSMC.... so again, i was asking could it be a process failure from TSMC.

Lastly, there hasn't been enough sample size of the CPU to say its Normal.
Please ask me how hard it is for to find a 5900X... please... i can tell you it was easier for me to get a 3090 and a 6900XT before i could even sniff out a 5900X not overpriced at 2x.

Also if we are already seeing this many failures on the web and reddit + posts even from PCMag that was what prompted me to post it.
Now extremetech:
And Techradar:
PCWorld:


When i get my 5900X i will definitely hold onto the box, but now i probably will wait on the board until i see what happens with the USB issue on the X570 i was going to pair it with along with a 6900XT i currently have on hand for SAM feature.

But now thats all on hold again....
 
Last edited:

biostud

Lifer
Feb 27, 2003
18,193
4,674
136
Since I have the same problem now with a b450 an 3800x, but did not have it initially. I suspect it is a bug in the AGESA code that supports 5xxx CPUs, as I have updated the bios whenever a new version was available, and the problems did start around the time I updated to the AGESA that allows 5xxx to run on the B450 board.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
Where do you think the PCI-E lanes come from?
From a CPU with normal RMA rate verified by multiple system vendors across the world.

Now extremetech:
based on the PowerGPU Tweet
And Techradar:
based on the PowerGPU Tweet
based on the PowerGPU Tweet


At least Gordon from PCWorld is a veteran journalist who knows why doing your homework is important:
The first vendor said it hasn’t seen any issues with its Ryzen-based systems. The company closely monitors its build line and RMAs (return merchandise authorizations). If a component surpasses a two-percent failure threshold, it’s immediately flagged.

A second vendor echoed the first. Nothing unusual has been reported on its quality control, RMA, and tech support lines related to Ryzen 5000 CPUs, it told PCWorld.

A third vendor provided even more information. The company said it isn’t seeing PowerGPU’s reported failure rates with its own systems. Interestingly, however, the vendor actually shared data indicating that Ryzen parts are failing the company’s internal quality screening at a higher rate compared to Intel chips—almost three times as high:
  • Ryzen 5000 series fails at 2.9 percent.
  • Ryzen 3000 series fails at 3 percent.
  • ThreadRipper 3000 series fails at 2.5 percent.
For comparison, the company' data on Intel chips:
  • Intel 9th-gen fails at 0.9 percent.
  • Intel 10th-gen fails at 1.2 percent.
Let’s define ‘fail.’ The vendor noted that like most PC vendors, it still ships more Intel-based CPUs than AMD-based CPUs. The sample size affects the data: If you bake one dozen cookies and one burns, that extra-crispy one's going to stick out more than if you bake five dozen cookies and a few burn.

The third vendor torture-tests systems with fully loaded memory slots—even if they’re not eventually sold that way. The vendor also explained that its failed CPUs aren’t dead, as PowerGPU reported—they just don’t pass its requirements to run low-latency or high-frequency RAM fully loaded.

Interestingly, the third vendor also said there was no apparent rhyme or reason to CPUs that didn’t pass muster. For example, it found more Core i9-10900K chips passed its tests than Core i9-10850K chips, while the 64-core ThreadRipper 3990X had more success than the 32-core and 24-core versions. The third vendor said it wouldn’t describe its particular rejection rates as a “problem.”

Is there any chance we wait for at least one more system vendor to complain about AMD failure rates before we make this take off?
 

crashtech

Lifer
Jan 4, 2013
10,521
2,111
146
What is complete and utter failure is inability of AMD to provide 5950X in any quantity in most of Europe. Supply is so bad, that vendors have not seen any 5950x yet. Can't really comment on failure rates when you've seen zero CPUs :)
Only Europe? They are virtual unobtanium in the US also, afaict. I do see them on ebay.de if you want to pay 980+ euros.