Re: Asus announces 'immediate internal review' of 800-series motherboards following string of 9800X3D failures ....

On Mon, 1/26/2026 11:07 AM, VanguardLH wrote:
> "Mr. Man-wai Chang" <[email protected]> wrote:
> 
>> Asus announces 'immediate internal review' of 800-series motherboards 
>> following string of 9800X3D failures — users report multiple chip 
>> failures in recent days | Tom's Hardware
>> <https://www.tomshardware.com/pc-components/motherboards/asus-announces-immediate-internal-review-of-800-series-motherboards-following-string-of-9800x3d-failures-users-report-multiple-chip-failures-in-recent-days>
> (January 23, 2026)
> 
> ""We are aware of recent reports concerning AMD Ryzen™ 7 9800X3D CPUs
> and ASUS AMD 800-series motherboards, and we have initiated an immediate
> internal review," ...
> 
> "While the 9800X3D has been the victim of multiple reports of failures,
> these have previously been largely confined to ASRock motherboards."
> 
> I have an Asrock Z390 Taichi mobo, but Intel, not AMD.  Asrock is a
> spinoff from ASUS.  Got it back in April 2019.  I switched on Secure
> Boot, but ran into software using unsigned drivers, and Secure Boot
> doesn't like that.  When I disabled Secure Boot, the mobo wouldn't boot
> anymore.  Couldn't even get to the POST screen.  Couldn't burn in the
> lastest BIOS firmware, either.  Wouldn't boot that far to read drives to
> load the flash program.  After working on it for days, and almost 3 days
> with Asrock support, they decided to swap the motherboard.  The software
> dev eventually signed their driver, so I could enable Secure Boot, but
> if I do then I won't be disabling it.  Never bothered flashing to later
> firmware, either, but the replacement mobo did have a later version.
> 
> The article doesn't mention Secure Boot, but it isn't clear just what
> were the failures.  It mentions ASUS suggests burning in new firmware
> for BIOS, so maybe some of their code was incompatible with the AMD
> Ryzen 7 9800X3D which came out this January.  However, the article says
> ASUS is looking into all their 800 series mobos (which sounds like
> they're discussing the chipset, not CPU), and those came out in 2009.
> 
> Wikipedia also has their own blurb on the failures:
> 
> https://en.wikipedia.org/wiki/ASRock#Ryzen_9000_CPU_failures_on_AM5_motherboards
> 
> mentioning the failures were known back to Feb 2025, a year before the 
> Tom's Hardware article.  All of that started 6 years after my Asrock
> failure with Secure Boot, so the boot failures noted in the article may
> be due to other firmware code in BIOS, and maybe only regarding AMD.
> 
> Since the BIOS firmware probably comes from American Megatrends (AMI),
> wonder if the fault lies with ASUS/Asrock or AMI.
> 
On modern motherboards, the "chipset" has little to do with the CPU.
For example, some AMD motherboards use a chip from Asmedia as their
Southbridge, and it is more or less connected via PCI Express lanes.
It's not like there is a "custom level interface" that marries
the chip. In fact, someone bolted one of those Southbridges
to a PCI Express plugin card to make a six port SATA interface, which
shows how generic the function is.

The Wiki article mentions a PBO settings issue (Precision Boost Overdrive)
as a possible root cause. Which is a low-hanging fruit as an excuse
for a failure. I expect there is more to it than that. After the
number of incidents involving the same sorts of issues, this is
going to be reviewed pretty carefully before any motherboard
is released.

But in terms of companies and their BIOS team, Asrock seems
to release an awful lot of BIOS releases, for little apparent gain.

Asus has more engineers to throw at the problem, than Asrock would have.

The 9800X3D is an AM5 chip, and according to Wiki, AM5 was
released on "September 27, 2022", so any motherboards being
reviewed would be boards after that date. You can see the kinds
of Southbridges that would be under review (the Southbridge having
nothing whatsoever to do with the failures, but the Southbridge
is tied to the "generation" or "design epoch" of the failures).

https://en.wikipedia.org/wiki/Socket_AM5#800_series

    B840 B850 X870 X870E also daisy chained 2 "Promontory 21" silicon as X670/X670E.

The Southbridges are generic enough, you can cascade them. Which
means they have PCI Express lanes on the downstream side for the purpose.
When you have two Southbridges on AMD, they go under a large heatsink
and you may end up with a 40mm fan mounted on it. When I bought an
AMD motherboard, I made sure to buy a motherboard with only
the one-chip-Southbridge solution, so it could be convection cooled.
One chip might be seven watts, two chips fifteen watts, and having
to enable the downstream PCIe output likely costs a watt or two.

          "IN" "OUT"
        --- X870E --- X870E --/

   Paul