AMD Blames Motherboard Makers For Burnt-Out CPUs (arstechnica.com) 35
An anonymous reader quotes a report from Ars Technica: AMD's X3D-series Ryzen chips have become popular with PC gamers because games in particular happen to benefit disproportionately from the chips' extra 64MB of L3 cache memory. But that extra memory occasionally comes with extra headaches. Not long after they were released earlier this year, some early adopters started having problems with their CPUs, ranging from failure to boot to actual physical scorching and burnout -- the problems were particularly common for users of the 9800X3D processor in ASRock motherboards, and one Reddit thread currently records 157 incidents of failure for that CPU model across various ASRock boards.
In an interview with the Korean language website Quasar Zone (via Tom's Hardware), AMD executives David McAfee and Travis Kirsch acknowledged the problems and pointed to the most likely culprit: motherboard makers who don't follow AMD's recommended specifications. Some manufacturers have historically shipped their AMD and Intel motherboards with elevated default power settings in the interest of squeezing a bit more performance out of the chips -- but those adjustments can also cause problems in some cases, especially for higher-end CPUs.
In an interview with the Korean language website Quasar Zone (via Tom's Hardware), AMD executives David McAfee and Travis Kirsch acknowledged the problems and pointed to the most likely culprit: motherboard makers who don't follow AMD's recommended specifications. Some manufacturers have historically shipped their AMD and Intel motherboards with elevated default power settings in the interest of squeezing a bit more performance out of the chips -- but those adjustments can also cause problems in some cases, especially for higher-end CPUs.
"disproportionately"? (Score:3)
Making the CPU attractive to gamers by juicing up the level 3 cache was probably an intentional design + marketing decision.
Re:"disproportionately"? (Score:5, Interesting)
Re:"disproportionately"? (Score:5, Informative)
Re: (Score:3)
Exactly. Intel entirely at fault for their part - and repeatedly alternating between deflecting and providing multiple downgrading "fixes" while somehow managing to avoid false advertising blowback.
Re: (Score:1)
Re: (Score:2)
Those type of gamers are likely going to tweak the voltage regardless. So perhaps the slightly higher default voltage plus an additional tweak is sometimes enough to tip things over.
Re: (Score:2)
Hmmm (Score:4, Interesting)
Re: (Score:3)
Have you tried a slight decrease in voltage to see if that "cures" the problem?
Re: (Score:2)
Could also be a memory issue. I've found that sometimes the XMP profile for RAM doesn't have quite enough voltage to work reliably. After increasing the voltage by 0.05V the system became stable.
Re: (Score:1)
Sounds like DDR5 training. Is the kit you are using on the QVL list for the motherboard that you have?
I learned this after many years of just slapping whatever kit I liked in whatever board I liked. The QVL is the qualified vendor list, and it has a list of validated ram models that will work well. You can find it on the support section for your motherboard.
A great way to tell is if your system memory is no longer using the XMP/EXPO timing/frequency. Like, if you bought 5600 and set the EXPO profile, and no
Re: (Score:2)
The QVL list is a comfort blanket. Most vendors publish lists so specific that it's actually hard to get the RAM even if you're looking for it. The reality though is vendors are limited in what they can comfortably and quickly test, and often limited to baseline performance devices rather than any cutting edge stuff. "Slapping whatever kit" is still a valid way of building a PC.
But you are right in that it may be the cause, the solution though is to try different RAM, not necessarily run back to the QVL lis
Re: (Score:1)
Oh, well that's a nice to know. I had heard that the best way was QVL. Maybe sticking with vendors that are commonly QVL'd is valid, but you could write a book with all the stuff I don't know.
Most of the time the QVL is a distant afterthought. There really is something to be said about boutique PC vendors, systems integrators, and even apple, beyond, oww my wallet! As I get older, I really AM starting to get why people want to buy and not build. I been building for 25+ years. Sometimes you just want somethi
Re: (Score:2)
Having seen what people get when they buy is precisely what is keeping me in the building game even when I can afford to pay someone to do that.
The issue is in the performance side. It's one thing to get just a computer, HP, Dell, Lenovo, anyone of those can fit that bill. But when you get into performance side of things you start seeing some real incompetence across a wide range of system integrators delivering stuff that is broken, misconfigured (e.g. putting in a fancy stick of DDR and not setting up the
Re: Hmmm (Score:2)
ASRock ? (Score:2)
Really? People still buy their garbage? Back when I was still building systems for myself and customers, I quickly learned to stay away from this brand. I see they haven't changed.
Re: (Score:2)
My ASRock motherboard is doing just fine. But I am not a gamer. I don't even have a video card. The Ryzen 4600G does just fine.
Re: (Score:2)
What brand do you recommend? The thing is, people with experience on a brand will literally all be able to cite bad experiences. In my experience precisely zero companies are immune from producing rubbish.
MSI has horrendous BIOS implementation. It's just shitty all around to the point where I remember core features of the original Zen processor not being implemented. Even now they suck when it comes to pushing RAM timings.
I've sent multiple Gigabyte boards through RMA (but still use Gigabyte now personally)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Every manufacturer has their moments.
I have been a pretty ardent fan of Asus motherboards. But I have definitely had some bad boards by them. Overall though, they have been good.
My own rule of thumb when buying a motherboard (since I don't follow hardware very closely anymore) is to look at the cheapest board available that supports the CPU and then buy something with brand recognition in the 3x price range.
Re: (Score:2)
The X3D parts are particularly fragile.
"makers who don't follow AMD's recommendations" (Score:2)
You mean they didn't read the manual?
Re: (Score:2)
most manuals are write-only
This is the one thing I don't like about AMD (Score:1)
Nvidia and Intel both do that. They leave quite a bit of headroom for manufacturers to either run things out of spec or cheap out on electrical components.
A huge part of the problems AMD has on their video cards for example is manufacture
Re: This is the one thing I don't like about AMD (Score:2)
As for Asrock? They probably just had a misunderstanding during the design, or used a preliminary specsheet that was later updated. These things happ
Re: (Score:2)
It's Asrock (Score:1)
I had a potentially related problem (Score:2)
I tried the new CPU in the Gigaboard and the system started right up. I put the old CPU back in... and the system still failed.
I tried the old "bad" CPU in the other system with an Asus board... and it was fine.
Why would one CPU fail in one board but
Lose Sockets? (Score:2)