Core i9-13900KF RMA unit 2: The second RMA unit, or the RMA of the RMA had a promising start. Following the eTVB fix, it was stable in most of the games we tested. Until The First Descendant descended upon it: Crashes every other minute were the highlight of the launch. It took me two days to get through the tutorial which otherwise takes 15 minutes. Here’s an Event Viewer log containing all 44,242 error events.
I wonder how many users will run The First Descendant and then ask for a warranty replacement? I mean, savvy users will just keep giving error logs to Intel until they finally get a sample that is stableI still don't think a recall is necessary. If your CPU encounters the issues then Intel will replace it under warranty.
For months, today included, I continue to see users only now becoming aware of this issue. And reacting to the news finally knowing what is wrong with their system. Many many posts about endless troubleshooting, blaming the game devs, Nvidia drivers, failing storage, on and on. Only to learn later it was the CPU the whole time.Most people using 13/14900K haven't reported issues.
I saw a post this morning where the guy was waiting to see if his 5th RMA would get approved.I wonder how many users will run The First Descendant and then ask for a warranty replacement? I mean, savvy users will just keep giving error logs to Intel until they finally get a sample that is stable
Yeah and who knows what has been happening in third world countries where the shopkeepers can be pretty rude, especially if it's a big, successful shop.So I think stating most have reported no issues is a terrible metric to rely on at this stage.
This specific pie chart doesn't say much unless it is normalised per SKU.Following image needs to be quoted big:
Note: Those are crashes by all CPU manufactures and chip generations.
And addendum: That forum post is one example where updating to the latest BIOS seems to have helped resolve the issue (though I'd personally not hold my breath due to Intel's lacklustre messaging).
I agree on the recall part, as long as Intel comes out with a resolution (cause+fix). I'm no longer convinced this isn't widespread though, the telemetry data was a shock (especially as it was partially confirmed by a source inside a big OEM).I still don't think a recall is necessary. If your CPU encounters the issues then Intel will replace it under warranty.
Most people using 13/14900K haven't reported issues.
NVM, misunderstood your comment.This specific pie chart doesn't say much unless it is normalised per SKU.
In the case of the game server Raptor Lake CPUs running in the W680 boards, I don't think they ever ran at "too high" frequencies in those setups. Certainly not compared to similar CPUs in desktop boards.
Ah, sorry, I thought that was the line for Alder Lake vs Raptor Lake. Not sure where I read that though.Amusingly the 13500 is actually Alder Lake silicon. So it's probably fine.
Some juicy quotes, adding more to this post as I listen.It's a 2.5hr conversation and Q&A but Wendell and the Doc get into the stability problems -
At least 50%, maybe 75% of the CPUs I have identified that something ain't right with them, if you were using them in a desktop use case, you probably wouldn't know. (42:50)
83C max logged temperature in one of the server populations (45:20)
I'm like 85% sure that at least a double-digit percent of the population is degraded. (47:00)
One support ticket complained about the game only working in compatibility mode. Wendel says it was CPU degradation, and the reason it worked in compatibility mode was because AVX gets disabled. (52:30)
Some of the gaming companies that have done the reporting are now mitigating denial of service attacks. (1:03:50)
Some juicy quotes, adding more to this post as I listen.
It's more like 30-40 minutes, starting around 33:20. The rest of the stream is them talking about QC, NPUs, RISC etc.Appreciate the TL;DW cliffnotes. 2.5 hours is a lot of time investment.
Yes, Nvidia handled that about badly is possible (for consumers - it was "okay" for them except they never got any Apple business afterwards). And like this, the media barely covered it - still convinced that SA's Charlie D. constant digging made him persona non-grata with Nvidia and their astrosurfers.I agree on the recall part, as long as Intel comes out with a resolution (cause+fix). I'm no longer convinced this isn't widespread though, the telemetry data was a shock (especially as it was partially confirmed by a source inside a big OEM).
I just hope this does not end like Nvidia's bumpgate, because that was a textbook example of a company not taking (full) responsibility and attempting to pass some of the financial burden on consumers. For DIY products replacing the defective part is a reletively minor hassle, but for laptops it gets nasty fast.
Didn't realize this was still going on. I've heard so many people say they don't have an issue with their 13 and 14th gen chips, however I am seeing a lot of them for sale in secondary markets on other forums.
I don't see why they wouldn't in single or lightly threaded loads. . .
Caveat emptor.
W680 would be less likely to hit TVB clocks, I would think. I could be wrong. And again Wendel's initial video claims that the W680 samples didn't exceed 5.3 GHz.
It would seem from these last few posts, that if you never stress the CPU very much, it will survive. The comment about desktops, not gaming is the comments that mostly drives this overservation.Caveat emptor.
W680 would be less likely to hit TVB clocks, I would think. I could be wrong. And again Wendel's initial video claims that the W680 samples didn't exceed 5.3 GHz.
Not from what I remember in the video, no. I can rewatch it I guess . . .I thought that 5.3 GHz was the manually set frequency they used to mitigate the problem in machines they identified as having issues?
Not from what I remember in the video, no. I can rewatch it I guess . . .
more about stressing multiple cores than ghz ??If they never went above 5.3 GHz then I really don’t know what the heck could be going on.