AMD Ryzen 5000 ‘Zen 3’ Desktop CPUs & X570 Motherboards Have High Failure Rates, Reports PowerGPU

fightingfi said:
https://wccftech.com/amd-ryzen-5000-zen-3-desktop-cpus-x570-motherboards-high-failure-rates/


errrrrr are these guys PAID by someone who isnt A M D ............................? MY x570 and 3700x not a single issue here ever or my dads b450 2700x or my older 2700x x470 ever............whats your facts?
Click to expand...
In truth I have not heard much from my contacts OR experienced any large influx of Faulty AMD mainboards OR CPUs.

However I would imagine that AMD is more than likely stretching itself quite thin, and some 'relaxing' of the QC and binning process is almost guaranteed....
 
I would say it is bogus. The way the internet is quick to overblown everything now days. There really haven't been any reports outside certain boards having issues.
 
I doubt a small custom PC builder reseller would lie about something like that (an experience on such a extra tiny sample size made on a 1,600 follower twitter account....), but if the usually failure rate is 1% of 1%, that high occurance on small sample would still be almost impossible to happen if there was not a problem.
 
It would seem that 7nm production process is still not mature enough and it feels strange that they don't have a test bench to test that the processor actually boots up or what ever bios level test that it is reporting correctly. There will be huge problems, when chinese server farms become updating their Zen 2 Epycs > Zen3 Epycs this spring. There is a maximum of 8 CCRs to get up to 64 cores and if they have so many problems already with one CCR, then there is going to be HUGE uproar from their big server farm customers!
 
To me it is mind buggling how they manage to keep the air so pure, when they are UV burning the wafer in the production process. Typical bacterium is between 500 - 5000 nm, which is pretty small, when you try to filter it out from the dirty replacement air coming from the outside. There are multiple layers, where they place an insulation layer and new silicone layer ready for another UV burn and somehow there must be also the connective layer between the two consequtive silicon layers. MIND BOGGLING! If the human interaction with robots can be done remotely, then it would probably be best to create a complete vacuum for the clean process, because then gravity would take care of the small particle contamination issue.
 
Last edited:
Hi,
I can only add my experience. The 5900X I bought in late Dec 2020 did appear to be faulty and I had to send it back. The Ryzen 9 5950X that I bought in late Jan 2021 worked fine out of the box. I may have also had a faulty X570 Tomahawk though with the MB I can't say I couldn't have caused the problem with attempts to flash the BIOS repeatedly. In any case, the 5900X was the only CPU I've ever had that I could not get to POST in about 24-25 years of building my own systems. CPU failures never even entered my mind before so when I couldn't get this system to POST I went nuts over nearly a week replacing almost every component I could to isolate the problem. And yes, as mentioned below I looked for bent pins repeatedly. With only 2 Ryzen 5000 processors to test, I can't really conclude anything about their failure rates. Still it was a frustrating experience.

LT.

P.S. For anyone interested more detail is below.

After nearly 6 weeks of trying to get a non-scalper 5900X I did put in a successful order through Amazon on Christmas eve. While the original shipping date was mid-Feb 2021, the 5900X arrived 7 days later. It all happened so fast I didn't even have memory, a power supply or MB when I was given a revised shipping date from Amazon. SO I immediately ordered what memory I could, a PS and MB. Essentially getting what I could that was actually in stock. The following weekend I tried to get the system to post with the chip, cooler, 1 stick of memory and an old video card. I was hoping a Ryzen 5000 compatible BIOS was already on the Tomahawk. But there was no POST. The Tomahawk board has error LEDS, and my board never made it past the CPU LED. I spent the next 3-4 nights doing FLASHBACK after FLASHBACK with several USB2 sticks (2-4 GB each) formatted as FAT or FAT32. I watched several videos on how to flashback the bios and the board did seem to do everything I saw in the videos including the reboot. But each time there was no successful POST. I checked the processor about a 1/2 dozen times with a magnifying glass and with a camera to look for bent pins. I even recruited my youngest with better vision to look for bent pins. Nothing.

As a complete idiot I only remembered I had built a Ryzen 5 3600 system in late 2019 a couple of days in to this ordeal. In late 2019 I put together a compact computer with an ITX board and a small case as a 2nd family computer. For some reason I had thought I built an Intel i3 based system. But when I opened it up to look for a different brand of memory to put on the Tomahawk I noticed it had an AMD cooler. I checked the BIOS and I had a Ryzen 3600 processor! I didn't even have to use the USB Flashback system.

So I put the 3600 chip in the Tomahawk with high hopes that I could flash the BIOS and move on. But the Tomahawk didn't post with the Ryzen 3600 either. It would just sit there with the motherboard fan running, and the CPU LED staying on. Power did get to the CPU fans and the Video card fans would spin up too. I just couldn't POST. I tried 2 different video cards (Nvidia 760 and 1080Ti.) I tried both my Ballistix memory trying each of the DIMMS individually and using the 2nd system's G.Skill DDR-3200 memory too. Nothing.


That actually made me feel better, thinking the Tomahawk was faulty. So I ordered a MSI Unify board and that came a few days later. This time I put the 3600 on it first and it booted up right away. Everything in the BIOS looked correct. I checked the BIOS and it was the Nov 2020 BIOS which should have been the 1st compatible with Ryzen 5000 series. So I replaced the 3600 with the 5900X and AGAIN IT WOULD NOT POST! So I put the 3600 back on, and it POSTED. I then updated the BIOS to the latest, rebooted, cleared the CMOS, rebooted again and loaded all optimized defaults. Put the 5900X back on and NO POST. So now, I cleared the CMOS by removing the battery (MB already unplugged) for 10-15 minutes. Reconnected power and it still would not POST.

So I checked for bent pins AGAIN for hours. None bent. I called Amazon hoping I could ask for an exchange since it was such a chore to get the 1st chip and I don't live within hundreds of miles of a Microcenter to try there. The CSR said due to low stock and not knowing when they will get more, all they could do was offer a refund. I took that offer. The CSR did say that their policy was to check for damage and if they found any there would be a restocking fee. I guess he implied that I damaged the chip. I sent the chip back and it did take Amazon 3+ weeks to give me my refund though I did get the full refund so I guess they saw no damage either.

That put me back in line to get a new CPU. I used the Fairgame bot for about 3 weeks and set it to look for either the 5900X or a 5950X. I didn't want the 5950X as it was overkill for me (in many ways the 5900X would be too but these days I keep systems for 4-5 years). Still by this time I had every component but the chip (and a graphics card!, lol). So I was going to get anything that I could. On 1/31/2021 the bot snagged a 5950X from Amazon at retail prices. It too was supposed to ship many weeks later, but again I got an updated shipping notice about 4 days later and received the 5950X on 2/5/2021.

I popped it in the Unify (with the latest bios that I updated using the 3600) and it POSTED immediately! It was just like every system that I have built before the 5900X. I've only had it for a little over a week but it has been fine. I've been running it at stock except for enabling XMP in the BIOS to get my memory to run at DDR-3600. So far so good. It appears stable. I've spent a week testing and stressing it and the memory with no issues. In a week or two I might try PBO2 and a curve optimizer though after checking the thread about 5900 overclocking here on [H] I'm not sure it will really be worth it.

So this upgrade has been a pain in so many ways. Every component I have wanted has either been out of stock, or prices much higher than expected or both that its taken me 3+ months to get a system running. In the past, once I had a parts list, I was usually up and running 7-10 days later just accounting for shipping times. And I've never had a faulty CPU in my life and I've been building systems every 2-5 years (slower since Sandy Bridge) since around 1996. It felt weird sending a CPU back and worse knowing how hard it was to even get a 5900X.

LT.
 
That link actually says they're DOA, not just faulty. It's bullshit. If as many as 16% of Zen3 CPUs were DOA we would know about it. Hell, if 1% were DOA we would know.

Lots of people are having problems with Zen3 that may or may not be related to faulty CPUs. There's an absolute ton of data supporting Zen3 is a problematic launch. AMD screwed up Zen3 big time, to be crystal clear, largely with their shitty QA on the launch AGESA. But they didn't ship 8-16% of the CPUs DOA.
 
I agree that this report is bogus. With how the internet is so fast to jump on something this is the first we hearing about it. There would be a lot of nerd rage and Intel fanboys out in full force if the DOA was so high.
 
Not good for all people buying from scalpers, I hope getting the warranty working is not a hassle for them
 
I had a DOA 5900X. Booted up one time, got a WHEA error BSOD after two minutes, never booted again after lots of troubleshooting including a replacement MB. Exchanged at Microcenter and the new CPU is fine. There were plenty of similar reports on Reddit when I was researching what could be wrong.
 
I haven't seen many DOA reports elsewhere. Lots of WHEA problems, etc, though. I'm sure there are some, just not 8-16%.
 
I would want to see reports from large system builders like IBuyPower, CyberPower, etc. I've never even heard of these guys and doesn't look like they are that big. The sample size is rather small.
 
In my experience, DOA hardware is really rare - and I worked for 3 years as a reliablity engineer at a major semiconductor manufacturer. DOA hardware would indicate basic test coverage issues, burn-in issues (dialed back too quickly) or static discharge issues during packaging/handling.
 
schizo said:
That link actually says they're DOA, not just faulty. It's bullshit. If as many as 16% of Zen3 CPUs were DOA we would know about it. Hell, if 1% were DOA we would know.
Click to expand...
Except if PC builder are receiving different batch than usual customer (that did not go through Q&A expecting that the pc builder will do is own), purely speculating on my part. And that they received a particularly bad batch, i.e. that the numbers is ridiculously high say half a % but that they received an particularly bad batch.
 
Sure, I guess that's plausible enough. They got a pallet of CPUs from a single bad run, etc.
 
You must log in or register to reply here.
Back
Top