That's why I like Apple's vertical integration. If something breaks it's 100% their fault that they need to make right to the consumer, they can't tell you to GTFO and go blame someone else to escape responsibility.
Not a huge fan of that headline. It's completely plausible that motherboard vendors aren't following AMD's guidance and push the chips past their documented limits. It wouldn't be the first time, and it won't be the last.
The problem with Intel blaming 13th and 14th gen failures on motherboard manufacturers was that it happened even on motherboards which followed Intel's guidance to a T. This Ars Technica article doesn't seem to even try to make the argument that motherboard manufacturers are innocent, so the comparison to 13th and 14th gen failures is completely unwarranted.
Further, the article doesn't even investigate whether or not AMD's claim is correct. They could've compared AMD's max voltage specifications to the default settings configured by various motherboards.
Well, seeing how buggy motherboards are in general, I rather believe AMD when they say they're ignoring yet another crucial detail.
My previous Gigabyte motherboard had unstable TSC, incorrect IOMMU configuration, broken PSP and CPPC. My current Asus board also has C-state weirdness and broken ASPM and it killed one CPU(!) on default settings. I won't even go into all the bugs I've encountered with Lenovo laptops, ACPI bugs, ASPM bugs, Pluton bugs and so on and on.
Hetzner recently underwent a major maintenance project to replace all motherboards for their Ryzen 7xxx servers. Dont know if they used ASRock for those or not
It’s normal to expect AMD to put out list of supported motherboards and also test their chips on it right ? Then they can make the statement that the motherboard is not in our recommended list. Why do the blame game
Having worked with a MOBO manufacturer, I would say that this is somewhat plausible though not necessarily the fault of the engineers. In my experience, both users and company administration, e.g. sales, CEOs have a lot to do with failures. An example, users are used to a particular MOBO brand, they want to try the new gadget-a video card usually, which requires much more power than the recommended power supply. They call tech support asking for an opinion, tech support talks to sales, sales will want to make the, well, sale. They typically did not ask the engineers for an opinion, as they are overseas, and instead query a direct manager, who will typically test the compatibility of the device and for a few days. When the device passes the test, they relate the good news to the sales department, who will then contact the user with the good news, an offer of a complete system. The user buys the system with the new video card, takes it for a ride. Three months later the user calls in reporting heating issues, tech support walks the user through some test, the user typically provides the results, tech support does not have an answer for the issue, they ask the user to send the unit in for testing and repair. Unit comes into the shop, board gets replaced, tested for three more days, passes all tests, gets shipped back to the user and issue pops up again, user gets irate, threatens to sue, nothing gets fixed. All because the bottom line is the $3000 dollars that was made in the sale. The issue eventually gets sent up the chain to engineers, they provide a BIOS fix, given to the user, user flashes. Issue is fixed, two years later, new Video Card comes in, user decides he needs that vid card, the never-ending cycle continues anew. I remember an instance where machines from a former employer was touted as a small low-heat server, really tiny, machine was made with substandard components, basically, cost-cutting at its best. Machines overheated, CPUs damaged, real shit show.
So yes, MOBO manufacturers, users bare equal responsibility, a chance to make lots of money & commission from the former, the need to be the first to try cutting edge technology without regards for proper safety, limitations and hardware compatibility, from the latter.
It seems that over the decades, people have failed to learn or neglected, Moore's Law's increasing physical limitations, because hey why not ;)
Eh, AMD did the same thing literally last gen. Telling us that moob manufacturers didn't follow spec while they did, it's just that the spec wasn't clear and amd changed it intel style right after it came out.
But it has to be said, that it in this specific instance it seems that only asorck is affected (and has been for months at this point thanks to GN investigation) so my bet is to just wait and see what happens.
13 comments
[ 4.3 ms ] story [ 35.5 ms ] threadhttps://news.ycombinator.com/item?id=45041743 ("GMP damaging Zen 5 CPUs? (gmplib.org)"—18 hours ago, 192 comments)
The problem with Intel blaming 13th and 14th gen failures on motherboard manufacturers was that it happened even on motherboards which followed Intel's guidance to a T. This Ars Technica article doesn't seem to even try to make the argument that motherboard manufacturers are innocent, so the comparison to 13th and 14th gen failures is completely unwarranted.
Further, the article doesn't even investigate whether or not AMD's claim is correct. They could've compared AMD's max voltage specifications to the default settings configured by various motherboards.
My previous Gigabyte motherboard had unstable TSC, incorrect IOMMU configuration, broken PSP and CPPC. My current Asus board also has C-state weirdness and broken ASPM and it killed one CPU(!) on default settings. I won't even go into all the bugs I've encountered with Lenovo laptops, ACPI bugs, ASPM bugs, Pluton bugs and so on and on.
https://docs.hetzner.com/robot/dedicated-server/general-info...
So yes, MOBO manufacturers, users bare equal responsibility, a chance to make lots of money & commission from the former, the need to be the first to try cutting edge technology without regards for proper safety, limitations and hardware compatibility, from the latter.
It seems that over the decades, people have failed to learn or neglected, Moore's Law's increasing physical limitations, because hey why not ;)
But it has to be said, that it in this specific instance it seems that only asorck is affected (and has been for months at this point thanks to GN investigation) so my bet is to just wait and see what happens.