9800X3D Failures/Deaths in the UK?

[Drei]

Just to double check, the safest path for me at this point, is to update the MOBO to latest BIOS straight away? Can I build everything, power up then update, or do I need to use the BIOS Flashback before I put the CPU in?

My ram is 6400 but happy to leave it at EXPO 6000 if I can, VSOC at 1.18, and not touch Precision Boost Overdrive...
Has anyone tried using the Precision Boost Overdrive with Curve Optimizer: between -30 and -15?

 

[Rroff]

Assuming ASRock's explanation is correct as to the root cause; PBO is on auto by default so there is the possibility of damage within the space of starting up and updating with the CPU in place, though a pain updating via Flashback without the CPU would eliminate that risk. (It is worth checking a guide that explains the implications of package power limits and design current limits in this context as well).

 

[welshrat]

Just to double check, the safest path for me at this point, is to update the MOBO to latest BIOS straight away? Can I build everything, power up then update, or do I need to use the BIOS Flashback before I put the CPU in?

My ram is 6400 but happy to leave it at EXPO 6000 if I can, VSOC at 1.18, and not touch Precision Boost Overdrive...
Has anyone tried using the Precision Boost Overdrive with Curve Optimizer: between -30 and -15?

I have -10 on the cores, and -50mv on the CPU (sorry forgot where) based on a video about setting up safe settings for the 9800X3D (will attempt to find it later)

Here is the video on some good settings to help mitigate the issue

 

[Rroff]

If ASRock's story is still the same that it was overly aggressive PBO, maybe those early samples you mentioned were particularly sensitive to whatever they had done.

There isn't enough data to do anything other than guess.

 

[humbug]

My 9800X3D and Nova board is still sitting on my table, waiting for my RTX 5080 pre-order, however I have heard of potential issues with the platform, with CPUs failing (even though there is a lot of speculation on CPU, Motherboard (ASRock), RAM or the actual cause, with some thinking it is enabling of EXPO or suspended mode S3 state in windows). The issues tends to present itself as booting issues, black screens and eventually the CPU light on motherboard comes ON.

I reached out to OCUK who have assured me that they are not aware of any such issues, so I am wondering if anyone here has experienced them.

There is a Reddit page dedicated to this:

https://www.reddit.com/r/ASRock/comments/1iui7lx/9800x3d_failuresdeaths_megathread/

Some Ryzen 7 9800X3D CPUs are allegedly failing prematurely — over 100 cases documented based on user feedback (Update)

The majority of the Ryzen 7 9800X3D had isssues on ASRock motherboards.

www.tomshardware.com

This is an extract of their latest failure list:

So he goes from an ASRock board which killed his CPU to another ASRock board to see if it fixes it.

All it takes is 5 minutes on Google to find out what the problem is, this apparently is a professional and he's completely clueless.

Don't use ASRock boards, a lot of them are junk. Use Gigabyte.

There isn't enough data to do anything other than guess.

Aside from them almost all of them being ASRock. 82% with 13% being Asus, 4% MSI and 1% Gigabyte. I think MSI and Gigabyte, perhaps also Asus given their marketshare being larger than the other 3 put together are just normal expected failure rates, not for ASRock considering the percentage share they have for this, bare in mind also that its a couple of hundred reports with millions of them sold.

There is plenty of data.

Again stop buying ASRock boards, or ASRock products, Lower end GPU's..... i had a lower end ASRock GPU just a few years ago and it was hilariously crap. It was so bad i gave it away as there was technically nothing wrong with it so couldn't RAM it but also so bad i couldn't bring myself to sell it to anyone for money get some money back on it.

Just get Gigabyte, seriously i've had my 5800X sitting in a B550 Aorus Eliete AX V2 for 5 years and its never put a foot wrong, i'm expecting at least another 5 years out of it with how perfectly it still works, it was also inexpensive at £160. Gigabyte have also never been involved with any of this burning up of Ryzen CPU's crap, Asus have.

Just do a little research to find good reliable products and then learn how to take good care of them, its actually easy, i have been running PC's for 30 years and have yet to have anything go wrong.

 

[welshrat]

So he goes from an ASRock board which killed his CPU to another ASRock board to see if it fixes it.

All it takes is 5 minutes on Google to find out what the problem is, this apparently is a professional and he's completely clueless.

Don't use ASRock boards, a lot of them are junk. Use Gigabyte.



Aside from them almost all of them being ASRock. 82% with 13% being Asus, 4% MSI and 1% Gigabyte. I think MSI and Gigabyte, perhaps also Asus given their marketshare being larger than the other 3 put together are just normal expected failure rates, not for ASRock considering the percentage share they have for this, bare in mind also that its a couple of hundred reports with millions of them sold.

There is plenty of data.

Again stop buying ASRock boards, or ASRock products, Lower end GPU's..... i had a lower end ASRock GPU just a few years ago and it was hilariously crap. It was so bad i gave it away as there was technically nothing wrong with it so couldn't RAM it but also so bad i couldn't bring myself to sell it to anyone for money get some money back on it.

Just get Gigabyte, seriously i've had my 5800X sitting in a B550 Aorus Eliete AX V2 for 5 years and its never put a foot wrong, i'm expecting at least another 5 years out of it with how perfectly it still works, it was also inexpensive at £160. Gigabyte have also never been involved with any of this burning up of Ryzen CPU's crap, Asus have.

Just do a little research to find good reliable products and then learn how to take good care of them, its actually easy, i have been running PC's for 30 years and have yet to have anything go wrong.

The problem is a little more difficult when it comes to options though. I bought an asrock nova for £400 as it has 5 nvme slots. 4 of which can be filled without any issue and then the 5th disables the bottom pcie. Hardly any boards offer this many full speed slots.

I have had zero issues with this board and many others are fine but I do understand the concern here and buying again I would get an MSI board as they are the next inline with good nvme lane support. Gigabyte have bad lane sharing on 870e boards.

 

[Rroff]

The problem is a little more difficult when it comes to options though. I bought an asrock nova for £400 as it has 5 nvme slots. 4 of which can be filled without any issue and then the 5th disables the bottom pcie. Hardly any boards offer this many full speed slots.

I have had zero issues with this board and many others are fine but I do understand the concern here and buying again I would get an MSI board as they are the next inline with good nvme lane support. Gigabyte have bad lane sharing on 870e boards.

PCI-e provisioning is pretty bad on everything one way or another both Intel and AMD and all the vendor brands - there are a lot of boards which have hidden (or not so hidden) caveats when it comes to populating PCI-e slots and NVME, one I think is quite bad is that some Gigabyte boards the 1x and/or 4x PCI-e slots sometimes will become unstable at speeds above Gen 3 if the other PCI-e or NVME slots are populated in certain ways, and Gigabyte likely knows this as those slots default to 3.0 despite supporting 4.0 in the BIOS while the other slots default to 4.0.

There are MSI boards as well which disable certain things depending on what slots are populated, etc.

In my opinion both Intel and AMD should be held to account over the PCI-e lane provisioning despite the average consumer likely not using half of it - there is absolutely no reason other than Intel cheaping out at the expense of the consumer why LGA1700 for example couldn't have supported 1x PCI-e 5.0 x4 gen 5.0 NVME as well as a full PCI-e 5.0 x16 slot at the same time.

 

[welshrat]

PCI-e provisioning is pretty bad on everything one way or another both Intel and AMD and all the vendor brands - there are a lot of boards which have hidden (or not so hidden) caveats when it comes to populating PCI-e slots and NVME, one I think is quite bad is that some Gigabyte boards the 1x and/or 4x PCI-e slots sometimes will become unstable at speeds above Gen 3 if the other PCI-e or NVME slots are populated in certain ways, and Gigabyte likely knows this as those slots default to 3.0 despite supporting 4.0 in the BIOS while the other slots default to 4.0.

There are MSI boards as well which disable certain things depending on what slots are populated, etc.

In my opinion both Intel and AMD should be held to account over the PCI-e lane provisioning despite the average consumer likely not using half of it - there is absolutely no reason other than Intel cheaping out at the expense of the consumer why LGA1700 for example couldn't have supported 1x PCI-e 5.0 x4 gen 5.0 NVME as well as a full PCI-e 5.0 x16 slot at the same time.

Yep. Hence me choosing the Nova which fingers crossed doesn't kill my CPU. I do wonder what AMD and asrock know that we don't as they must have some decent telemetry now. I guess it will be buried under a rock somewhere.

 

[Deluxe1]

Still seeing lots of reports about dying cpus on asrock boards.

 

[xtal191]

Has there been many on here you had one die? Is it mostly an American thing?

 

[Tetras]

Has there been many on here you had one die? Is it mostly an American thing?

I can recall two threads, so definitely not loads, if you think I'm a mega OCUK surfer.

 

[Rroff]

Still seeing lots of reports about dying cpus on asrock boards.

I wouldn't say loads but there is a steady trickle - I've seen 2 people so far today posting of CPUs dying in this manner on AsRock X870 boards (not just 9800X3D though it is the most common one). EDIT: As in the CPU died today rather than recently, one of them is on 3.16 BIOS the other on latest.

 

[humbug]

PCI-e provisioning is pretty bad on everything one way or another both Intel and AMD and all the vendor brands - there are a lot of boards which have hidden (or not so hidden) caveats when it comes to populating PCI-e slots and NVME, one I think is quite bad is that some Gigabyte boards the 1x and/or 4x PCI-e slots sometimes will become unstable at speeds above Gen 3 if the other PCI-e or NVME slots are populated in certain ways, and Gigabyte likely knows this as those slots default to 3.0 despite supporting 4.0 in the BIOS while the other slots default to 4.0.

There are MSI boards as well which disable certain things depending on what slots are populated, etc.

In my opinion both Intel and AMD should be held to account over the PCI-e lane provisioning despite the average consumer likely not using half of it - there is absolutely no reason other than Intel cheaping out at the expense of the consumer why LGA1700 for example couldn't have supported 1x PCI-e 5.0 x4 gen 5.0 NVME as well as a full PCI-e 5.0 x16 slot at the same time.

I have a Samsung 990 Pro With HS. https://www.samsung.com/uk/memory-storage/nvme-ssd/nvme-pcie-gen-4-2tb-mz-v9p2t0gw/ (Great Drive BTW fast as ****}

PCIe 4, benchmarks over 7000 MB/s, that is the maximum bandwidth for PCIe 4 X4, in the top slot nearest the CPU, the GPU is still PCIe 4 X16.

My CPU, Ryzen 5800X has 24 PCIe 4 lanes, 20 for the GPU + NVMe (That's 16 + 4) and 4 for the Chipset, that on my Motherboard breaks down to 4X PCIe 3 for a PCIe 3 NVMe with the rest for an assortment of Sata, USB 2, USB 3 and USB 10 Gb.

It does not have 3 or more NVMe slots because it does not pretend the CPU has enough PCIe lanes for that, it has all the IO the CPU can provide, those motherboards with 3 and more NVMe slots will start taking PCIe lanes from other IO if you populate them, this is normal, its not good but once you understand what is actually going on, as i just explained you will realise this is not 'a unequally Gigabyte problem' if they add more IO than the CPU has it starts stealing from other IO if you actually try to use them, Asus, MSI, ASRock, Gigabyte all do the same thing.

My second MVMe slot has a PCIe 3 NVMe in it because that's all it can do, everything always works at the speed it supposed to, because my IO does not exceed the PCIe lanes the CPU can provide, you want more? get a Threadripper....

 

[TrakaisKjems]

I have 2 pc builds with 9800x3d one is for me and other was gift for my son on his birthday .
My build.
9800x3d , nvidia 5080 , corsair vengeance 6000mhz , msi b650m gaming plus .
Sons build 9800x3d , asrock taichi 9070xt , Asrock b850m pro rs wifi , same ram .
Sons 9800x3d was bought from overclockers in march and it failed last weekend. He was playing night before and it was running fine next morning he turned it on and there was green light on on post , everything was running except no signal to monitor . I did update bios from 3.25 to 3.30 and it booted i could update drivers to gpu and do all windows updates with pc restarts and play some games but when i turned it off and then on it again failed to post with green light on mb . Then i tried everything removing ssd and ram and gpu and cmos reset. Nothing worked to post can’t even get in bios . Last resort was to put it in my setup and to check if its going to post unfortunately it starts and after that red yellow light and after for short time green light and then no lights and no signal to monitor .

 

[sastusbulbas]

My Nova has currently only seen light use with Linux utilizing a 7600x.

Thought I had heard the last of these issues.

As I don't own an Nvidia 5080 or 5090, I don't have a use for a 9800X3D as yet.

Would be great to see some sort of verified information on this as an ongoing problem or not.

 

[humbug]

Another one with an ASRock board....

Its seems like the board is overvolting the SoC, there are burn marks on CPU pads corresponding with the SoC voltage supply.

 

[Quarktran]

Built a system with a 9800X3D and X870E Nova. Running since November last year. No issues at all. Think they are a great board, with decent PCIe provisioning. I'd caveat by saying I've always manually configured voltages and settings, and even though I've had no issues, I am running the latest Bios. For the large amount of these boards and chips sold, particularly this combination, it really is a small number of reports. I have never given it much thought.
Hope I haven't jinxed myself now.

 

[welshrat]

Built a system with a 9800X3D and X870E Nova. Running since November last year. No issues at all. Think they are a great board, with decent PCIe provisioning. I'd caveat by saying I've always manually configured voltages and settings, and even though I've had no issues, I am running the latest Bios. For the large amount of these boards and chips sold, particularly this combination, it really is a small number of reports. I have never given it much thought.
Hope I haven't jinxed myself now.

Same. Running nearly 6 months now with manual vsoc of 1.18 and CPU offset -50mv. Have expo enabled and sleep disabled, it's probably the most stable system I have owned. But they shouldn't be burning CPUs with default settings and you have to imagine that they have enough data now to know what's wrong.