You can't draw that conclusion from that data. They state, point blank, that they adjust the Motherboard defaults themselves. Most people do not, and most people don't build their own machines.
The only thing that you can argue specifically from that data is that if you follow their guidance, failures follow a distribution not dissimilar from other CPUs. So, if you want a reliable 14th gen, either order from them, or, build your own with a new CPU and follow their guidance, or, validate that other manufacturers also follow their guidance.
They are just using Intel's recommended settings. They are not undervolting or anything like that.
A retailer reports a similar level of return rate: https://www.tomshardware.com/pc-compone ... -prior-gen
If you can get past the sensationalist headline, they report:
"According to data from Les Numeriques, only 1% of AMD processors were returned in 2020, while Intel had a 1.75% return rate then. So, if AMD’s return rate remained stable since then, we can extrapolate that the Raptor Lake chips have a return rate of 4% to 7% while Raptor Lake Refresh processors would have 3% to 5.25%. We should also note that these numbers only reflect return rates that went through the retailer channels, not those that went straight to Intel."
And from one of the original reports from RAD game tools (owned by Epic Games) from their Oodle compression SDK notes "Only a small fraction of those processors will exhibit this behavior."
On the other hand, the evidence we've seen that purport to show failure rates of 50-100% are is far less reliable. One of the main sources, and the one that really set this whole story off is a blog post of a developer from relatively unknown game studio, Alderon Games: https://alderongames.com/intel-crashes. He (Matt) claims that they are experiencing a 100% failure rate and nearly the same rate for their laptops. Apart from those numbers being far too implausible to be taken as general failure rates, his post is also quite aggressive, and it's notable that he hasn't updated it at all despite more info and a specific microcode addressing the problem being released. I can think of reasons why he might be motivated to make such a post - as an excuse for bugs in the game he works on, and/or for financial reasons by investing in a competitor (note the subreddit)
The more believable reports of high failure rates (like 30%) for these chips came from servers (running the W680 chipset for 14900k) with specific workloads involving single threads on turbo running constantly, like Minecraft servers. Hardly reflective of the vast majority of users usecases.
What their own data shows is that by following their guidance, Intel CPUs in 13th/14th gen are still higher than 12th gen. What their data does not show is the relative comparison of 13th/14th gen to 12th gen, or AMD, from motherboard manufacturers, or, the vast majority of PC builders that people actually buy from.
Cynical me thinks that they want to remain competitive in the space and so have to "sell" the idea that selling 13th/14th gen if fine, without making it too obvious that they are doing so. I suspect that they know that they have plenty of customer's who insist in Intel over AMD and they want those customers to feel confident. I don't doubt their data, but I do doubt that their data is representative of the industry wide picture. I also don't think that we're going to get that data because it would largely have to come from Intel and I think that it would do more damage than good.
Yes the failure rates reported by Puget are higher than 12th gen. They are also lower than 11th gen. It's a real issue, one that Intel has acknowledged for a while now. But it certainly has been massively overblown.
Puget is a very reputable company that published lots of benchmarks and such, and they've even recommended AMD over Intel in the past.
Statistics: Posted by NER — Sat Sep 28, 2024 4:55 am