Allegations of Unusually High Failure Rate Among Ryzen 5000 and 500 Series Motherboards by System Integrator

[agatong55]

How many of these are actually DOA and not actually a bios update? Im not saying AMD doesnt have any DOA products but how many of these customers are buying new 5000 CPU and not realizing that you will probably need a BIOS update to run them so they just return everything and say its a DOA?

[cj09beira]

 

7 hours ago, leadeater said:

Yea but we have a fair ways to go before that's going to be a problem on consumer platform, Xeon is 3647 and EPYC is 4094 neither of which have that bad mounting problems, they do however make a nice large target for old Linus Drop Tips over there to drop the CPU into while building the system 

isn't it lga 3647 that gave several youtubers problems, many of them having the system booted without all channels?

[leadeater]

15 minutes ago, cj09beira said:

 

isn't it lga 3647 that gave several youtubers problems, many of them having the system booted without all channels?

I thought that was EPYC, could be wrong.

[leadeater]

47 minutes ago, TVwazhere said:

RGB Build 2019 initially didn't POST because the CPU wasn't fully socketed. The retention lever was not pulled all the way back when the CPU was being inserted, meaning the CPU pins were partially exposed upon very close inspection under the heatsink (this was a system I had verified previously posted on my own)

Should have used a hammer, that'll fully seat the CPU in to the socket.

[MageTank]

I can't speak for this integrator in particular, but the integrator I work for just did a very small run of 200 systems with some Ryzen CPU's and had exactly 0 DOA CPU's. Granted, 200 is a pretty small sample size in the SI world, you'd still expect to see some failures if what they are claiming is true. We've seen far more failures on AIO pumps being DOA above all else.

 

I also want to point out that the instability issue people are complaining about with Ryzen out of the box is real, though it mostly depends on your board and the BIOS AGESA implementation. I don't buy into the whole "customed SI BIOS" thing that a lot of other SI's preach when they slap their logo's on and call it a day, but man, some of the BIOS's these people release is utter nonsense. Here are some of the common issues I see on a daily basis qualifying boards:

• USB ports mapped incorrectly
• Wrong number of Type C ports listed (enumeration faults)
• Scary levels of stock voltage & uncapped current limits (MSI, you get the most guilt here)
• Secure Boot straight up doesn't work (keys cannot be loaded, some keys may be loaded and cannot be cleared, the context menu breaks and auto-discards the moment you leave the Secure Boot menu, etc)
• fTPM doesn't function at all (common on new BIOS updates that bring new processor support, they always forget to add processor TPM along with it, yet ship these BIOS updates out anyways and SI's will use them without thinking twice)
• Beta AGESA firmware being used in WHQL BIOS updates (Nice try ASUS, you really almost pulled this one off, FCLK shenanigans and all)

That's just a short list. Prior to working this job, I wouldn't have paid any attention to any of this, so I now believe people when they say they ran into these wonky issues on brand new hardware. Part of my job is going back and forth with vendors until we get a BIOS that works right, then hope that a customer doesn't immediately "upgrade" it back to the nonsense that was the original, lol.

[Guest]

2 hours ago, leadeater said:

Well I can do one better, Slot 1. Now that would almost never have a mount failure 

 

But yea, been building computers since the dark ages i.e. Pentium/ Pentium Pro. Other than the above Pentium II all PGA all the way up to I think the largest pre-LGA being AMD 939/940, I don't remember any bad mounts on those. I've had a suspected bad mount on my LGA-2011 4930k but it's under a full cover mono block with hardline tubing and I can get it to boot with all 4 memory channels so I ain't gonna remount it, wayyyy too much effort lol.

 

My last DIY bulk building was LGA775, don't remember exactly what they were but it was around 2009, everything past that has been OEM HP/Dell/Acer and then the things I've personally built are just my gaming systems, family computers and server builds for my lab so easily below 20 in recent history.

 

With how PGA works though it should be very hard to get a bad mount, you'd have to be using a cooler with a rather bad mount design that allows significant overtightening to warp the board which would affect LGA just the same as well. Maybe modern coolers are more to blame than the sockets?

Slotted cpu? Would be like reseating memory which is commonly suggested.

 

Can think on those old Nintendo cartridges too I guess.

 

I think what mainly causes this issue is corrosion. Like brass and gold get a coating where its just a poor contact with just enough resistance. 

 

I have been running lidless processors since 2009 up to and including my 2700x. So I use custom mounting and have bent the board enough to create a no post issue. Should be less issue for OE coolers, but they do mount rather tightly.

 

Just in this forum, I've seen mention to reseat memory and processor on numerous occasions. So its a highly mentioned fix I do believe. And it never hurts to try I suppose.

[Bombastinator]

There were numerous and severe problems with slot1 even ignoring that it dealt with much slower chips that needed many fewer pins.   Slot1 was awful.  There’s a reason it was abandoned and companies went back to the previous system. “The old one was better than the new one” is a pretty harsh criticism.   Imagine how long the slot would have to be for 1700 pins.  Much longer than a pciex16 slot

[JamesTheFolf]

13 hours ago, huilun02 said:

Yes they can get the bad units replaced. Meanwhile everything seems normal in the DIY space

Yeah it would typically look that way because people are only buying one CPU, but system builders buy hundreds so they are always going to have more ones that are DOA.

[Dillpickle23422]

oh great. just what we need when there's already a shortage

[Bombastinator]

In any case the culprit needs to be tracked down.  Could be a bunch of things

[MageTank]

32 minutes ago, valdyrgramr said:

Well, ASRock typically doesn't beta test their bios much which can cause problems even with boards as high as taichis.  I know as I and others have suffered from this.  In ASUS's defense on the power thing it was more of an exaggeration on some of the higher end boards, like mine, with the power thing and Jay isn't always right.  I used a more accurate monitoring tool than the BIOS itself and HWMonitor which is HWiNFO64 to monitor it, and it's not as bad as Jay's exaggerations.  <.<  I could see Biostar having bios issues, MSI potentially, and Gigabyte I've never had issues with their bios. But, I tried one board of their's and it was DOA for Ryzen.

I am not speaking for anyone else. I am speaking about boards I have physically had my hands on and have worked with them to address in BIOS updates for my employer. I didn't watch Jay's video, I just know what I've mentioned from first hand experience has been par for the course as of late.

 

Also, ASRock isn't free from the list of issues I had, they just didn't get named because the issues they typically have are the same issues every board manufacturer suffers from (port mapping, secure boot, TPM, etc). Luckily, their boards didn't suffer from the same unlimited current settings or buggy AGESA firmware (AMD specific) so far. I will also say that of all the vendors I personally work with, ASRock has been the easiest to work with. The only vendor I can't speak for is Gigabyte as I have never worked with them professionally and have no contacts with them yet.

[Letgomyleghoe]

11 hours ago, LAwLz said:

municipality schools with laptops for their students

laptops and CPU's are far different, laptops should be booted up and tested before shipping whereas CPU's are most likely not. 

[MageTank]

18 minutes ago, valdyrgramr said:

Well, the x470 Taichi had a buggy AGESA with 3.90.  The point of the update was to increase the boot time into Windows 10.   However, in a lot of cases people would turn the system on, no post, and have to swap the PSU off then on or hit the restart button to turn on the case.   GG  My favorite response from them was to "rollback" when you can't.  They then sent me a tool that does nothing more than brick the board.   e.e  GG again ASRock.  So ya, I'm done with them.  I mean on one hand the Taichis have godly VRMs, but at the same time I'm not dealing with their support people ever again nor am I dealing with their SW people again.

 

With Gigabyte I can't stand their newer bios style, but their classic is fine.

MSI and ASUS I've had the best luck in the bios with with modern Intel and AMD.   However, MSI comes with some bad bloatware imo.

Never integrated with the X470 Taichi, so I can't speak about that board in particular. I can only share what I've found with the boards I've tested with. The USB port mapping issue has been common among all vendors I have worked with. If anyone has wondered why it takes time for their ports to enumerate when they plug a flash drive in, that is exactly why.

 

It's the weird things we often overlook when we build our own personal systems that can become a big problem if left unchecked for mass production.

[Bombastinator]

1 hour ago, Letgomyleghoe said:

laptops and CPU's are far different, laptops should be booted up and tested before shipping whereas CPU's are most likely not. 

They should do it to both.  It’s required for any kind of binning. 

[Kisai]

17 hours ago, Bombastinator said:

So they not go through testing or something?

Not every defect is found at the assembly point. For example of all the dell laptops I've done work on, the highest defect rate would be on 17" laptops, because there aren't that many, but every Precision 7720 has had mystery problems. Most 14" Latitudes, have battery swelling problems within a year. Precision 5530's are having batteries swell and popout the touchpad. These are relatively high defect rates that aren't going to be detected until the user has the item.

 

SI's test their products more extensively before they go out the door, and a 6% failure rate to me suggests a logistics failure on their part if other SI's are not reporting the same. Like I'd like to see what motherboards and cpu's they were using.

Quote

The failure on the new AMD CPUs are still too high.

AMD Ryzen 5950x x50 units 8 doa
AMD Ryzen 5900x x50 units 4 doa
AMD Ryzen 5800x x100 units 4 doa
AMD Ryzen 5600x x120 units 3 doa

Like what kind of bulk shipment material/handling is going on when they are bulk ordered from their supplier?

 

They've deleted the original tweet and replaced it with this one:

 

 

[Bombastinator]

5 minutes ago, Kisai said:

Not every defect is found at the assembly point. For example of all the dell laptops I've done work on, the highest defect rate would be on 17" laptops, because there aren't that many, but every Precision 7720 has had mystery problems. Most 14" Latitudes, have battery swelling problems within a year. Precision 5530's are having batteries swell and popout the touchpad. These are relatively high defect rates that aren't going to be detected until the user has the item.

 

SI's test their products more extensively before they go out the door, and a 6% failure rate to me suggests a logistics failure on their part if other SI's are not reporting the same. Like I'd like to see what motherboards and cpu's they were using.

Like what kind of bulk shipment material/handling is going on when they are bulk ordered from their supplier?

 

They've deleted the original tweet and replaced it with this one:

 

 

If batteries are swelling it’s because they’ve either got counterfeit parts or they’ve got the wrong specs they’re designing to.  Swelling batteries happen when the electrolyte gasifies the swelling is sort of a safety feature of the system design.  its stupendously dangerous, and should take a LOT longer than a year to happen.  Something is very wrong.

[Kisai]

15 minutes ago, Bombastinator said:

If batteries are swelling it’s because they’ve either got counterfeit parts or they’ve got the wrong specs they’re designing to.  Swelling batteries happen when the electrolyte gasifies the swelling is sort of a safety feature of the system design.  its stupendously dangerous, and should take a LOT longer than a year to happen.  Something is very wrong.

So are you suggesting that Dell is using garbage batteries in their business laptops? To be fair, not every single laptop is doing that within a year of purchase, as some of the laptops likely sat in a warehouse for 6 months before ever being used.

 

No, I think the problem is that Dell is under-engineering their ultralight laptops for people to use it as their primary computer for 8-12hr's per day.

 

 

[CD_5]

On 2/14/2021 at 8:05 PM, SPARTAN VI said:

Summary

System integrator PowerGPU reported that they had purchased an estimated 320 units of Ryzen 5000 processors for assembly in custom builds. Of those 320 CPUs, 19 of them were dead on arrival (DOA), which represents a near 6% failure rate among units shipped to them. They classify "DOA" as cosmetically perfect processor (e.g. no bent pins) that did not POST during their initial build, but would POST after swapping to a different but identical processor.

 

Quotes

 

My thoughts

Granted a single data point - one system integrator's experience - does not remotely establish an overall trend, I hope this leads to an open conversation from other system integrators and DIY builders who can corroborate and/or challenge PowerGPU's figures relative to their volume. We simply need more data for the law of large numbers to work. Currently, it is more likely that there's something environmentally isolated to PowerGPU's situation than just being "extremely unlucky." PowerGPU did state that prior to Ryzen 5000, they typically experience a failure rate of less than 1%

 

Sources

FPS Review article

WCCFTech article

I am seeing issues just as bad as doa, like faulty imc. The failure rate currenty only includes doa not any other major issue that cause it not to run properly. 

[Bombastinator]

15 hours ago, Kisai said:

So are you suggesting that Dell is using garbage batteries in their business laptops? To be fair, not every single laptop is doing that within a year of purchase, as some of the laptops likely sat in a warehouse for 6 months before ever being used.

 

No, I think the problem is that Dell is under-engineering their ultralight laptops for people to use it as their primary computer for 8-12hr's per day.

 

 

It’s the swelling.  Swelling is bad and is normally associated only with knock-off batteries.  There’s either not enough battery or the wrong kind.  Either way it’s a problem.  

[Fnige]

On 2/16/2021 at 12:24 AM, leadeater said:

Well I can do one better, Slot 1. Now that would almost never have a mount failure 

What would it be like if intel/whatever continued with using Slots for CPUs?

[Bombastinator]

36 minutes ago, SlimyPython said:

What would it be like if intel/whatever continued with using Slots for CPUs?

There were 2 big problems with slot1

A. It limited the space available for a cooler.. a lot.

B. The slot was basically a row of double pins.  As more pins are needed for a processor the slot has to get longer.  A slot for an lga1200 would be 600 pins long if a pin takes up half a mm (normally it’s more like 1mm) that’s a 300mm slot.  For people who thing in English measurement that’s just shy of a foot long. 

[Batman 007]

And, here I'm waiting for non-X Ryzen CPU..

[leadeater]

1 hour ago, SlimyPython said:

What would it be like if intel/whatever continued with using Slots for CPUs?

Hmm, probably horrific. Outside of server market everyone hated slot CPUs.

[leadeater]

57 minutes ago, Bombastinator said:

B. The slot was basically a row of double pins.  As more pins are needed for a processor the slot has to get longer.  A slot for an lga1200 would be 600 pins long if a pin takes up half a mm (normally it’s more like 1mm) that’s a 300mm slot.  For people who thing in English measurement that’s just shy of a foot long. 

They would have just moved to a high density Slot/Socket like which is used currently in servers for CPU/Memory expansion boards, problem there being is these high density slots/sockets have really small (although longer and stronger) pins in them anyway which really do make them not significantly different than an LGA CPU socket as we have now.

[Bombastinator]

Just now, leadeater said:

They would have just moved to a high density Slot/Socket like which is used currently in servers for CPU/Memory expansion boards, problem there being is this high density slots/sockets have really small (although longer and stronger) pins in them anyway which really do make them not significantly different than an LGA CPU socket as we have now.

Like agp?