It doesn’t have to affect every pc, it is affecting a significant proportion and in that it presents a significant risk.
What evidence do you have that supports a majority of affected parts are failing?
I didn’t say majority, I said proportion. In services where it matters, not necessarily gaming, it presents a risk that is significant.
So we go back to this same thing yet again: why should Intel spend billions of dollars on a recall that doesn't seem to warrant it?
Or Kia with something so bad as to "Don't park it indoors or it may burn your house down" but hey keep driving it for 10 months before we even issue a fix.
There may be abnormal degradation not that there is guaranteed to be.
Intel absolutely needs to do something with regards to this. While there won't be a guaranteed list of which parts are most likely affected they should be able to write software that pulls off of VID information to tell people whether or not they're likely to have had issues.
I would never buy a used CPU in the first place unless it was for a really old platform. There's no way to know what people have done to them, and with a situation like this one it's now doubly dangerous.
Unfortunately the warranties don't reflect this reality which companies use as a scapegoat. I've said in other posts here that I absolutely think Intel needs to extend the warranty on these parts. There's no way for an end user to determine the likelihood of damage unless it's already experiencing signs.
If someone's CPU fails due to this it's something Intel absolutely should cover (and as I said above I think they should extend the warranty). That doesn't mean Intel should be throwing away billions of dollars to rectify the situation when the majority don't appear to be negatively affected.
Because they would like people to trust them in the future? This would be a mild single year *profit* hit, let alone a loss. Take the pain, show your customers you go 'above and beyond' for YOUR mistakes.
You didn't click where it says "View All".
| Tier | 65W Mainstream | 35W Mainstream | Embedded | Low-power Embedded |
|---|---|---|---|---|
| i9 | i9-13900 | i9-13900E | i9-13900TE | |
| i7 | i7-13700 | i7-13700T | i7-13700E | i7-13700TE |
| i5 | i5-13400, i5-13500 | i5-13500T | i5-13400E, i5-13500E | i5-13500TE |
| i3 | i3-13100 | i3-13100T | i3-13100E | i3-13100TE |
Put simply, people put their faith in intel.
FWIW, disabling the E-cores has been tried with relatively little succes.
Another way to do this (at lower cost to Intel), is to extend the warranty period and relax the terms for claiming it. It has been suggested that retail boxed CPUs go from a 3 year warranty to 5, which I think is reasonable.
Yes but these attempts are on cores that have already been damaged. The Bartlett chips will never have E cores so if the issue is the variations between P and E make the bus have issues then those issues cannot happen without both types of cores.
Yes, and here the upcoming microcode update is the recall. Recalls don't happen immediately upon problem discovery, even in things as serious as automotive recalls: https://onlinelibrary.wiley.com/doi/10.1111/jscm.12160
This Intel issue has three out of four.
Only for new car sales, and that due to a federal law (US). Used cars can be, and are, sold when subject to recall. https://www.consumerreports.org/car...ll-guide-your-questions-answered-a1115780728/
I said only the i9 and i7 will lack E-cores, but it appears a version of the i5 will also use this die:
I'm not sure about that, since disabling the E-cores is said to have some benefit. On Alder Lake, disabling them enabled the ring bus to clock higher. On both Alder Lake and Raptor Lake, disabling them also seems to disable the associated slice of L3 cache.
The definition of Point-of-Sale is changing, in the era of self-checkout and smart shopping carts. Intel has a whole section of their website devoted to it:
No, that's not what I said -- all I said is that if the flaw is architectural (which frankly wouldn't surprise me because I don't expect that this kind of stuff is controlled in microcode by default) SPR could be affected too.
You are disregarding the factors that skew the statistics -- i9 owners are more likely to try to push the chip harder, after all they paid for the fastest chip. They are also more likely to complain than i5 and i7 owners. And there are non-K users who probably bought some prebuilt system and just shrug and reboot when it crashes instead of reporting.
If only there was some article where all this was already laid out for you to read:
You can't know that it isn't architectural unless Intel decides to tell us. They didn't explicitly spell out T series parts but they did say mainstream non-K models, and T series qualifies for that.
Again, no. RTFA.
Thanks for the heads up, in an ideal world those two readings would match. But on my Asus Prime Z690 P (w 13900kf), and my Aorus Z690i Ultra Lite (13600k) the external measured vcore*VR measured current matches the reported CPU power within a few watts, under load, across the frequency spectrum unless I really push the sliders weird. Even when the CPU requested volts is a couple to a few dozen mv less than the measured volts. Then the CPU requested volts* amps supplied comes out less than watts reported. But that is just on the two lower end Z boards I have. I can't speak for other motherboards.VID is how much the CPU requests, vcore is how much it actually gets from the mobo. The "goal" should be to make these 2 match, for starters, because that's the only way to get actual power usage from hwinfo. If vid and vcore don't match your power readings will be off. In order to make these 2 match you need to play around with AC / DC LL on advanced lite load settings. Setting up the correct values is based on motherboard's vrms mostly.
As you can see from my SS, min / avg / max vcore and VID are as close as possible. Maybe could tune a bit further to close that 0.003 difference but im too bored to do that
So Intel isn't lying, but maybe Intel is lying? You can't have it both ways like you're trying to here.
This is beyond absurd to suggest that somehow this is all because of purchaser psychology.
The voltage selection algorithm isn't just randomly applying voltages at random times. It's based on the voltage requests which only go up under load and boost. There's no debate over whether or not every CPU has this flaw, but rather whether or not it damages them all.
Here let me copy the actual Intel quote for you since you're too ridiculous to hunt it down yourself:
Base power on T series are 35W so...
I clicked, but only E and TE models are marked as IoT in the column "CPU Category".
| CPU Part Number | CPU Category | Processor Threads | Intel® Smart Cache (L3) | Processor Base Power | Single P-Core Turbo FreqB | Single E-Core Turbo FreqB | P-Core Base FreqB | E-Core Base FreqB | Graphics Execution Units (EUs) |
|---|---|---|---|---|---|---|---|---|---|
| Intel® Core™ i9-13900E | IoT | 32 | 36 MB | 65W | Up to 5.2 GHz | Up to 4.0 GHz | 1.8 GHz | 1.3 GHz | 32 EU |
| Intel® Core™ i9-13900TE | IoT | 32 | 36 MB | 35W | Up to 5.0 GHz | Up to 3.9 GHz | 1.0 GHz | 0.8 GHz | 32 EU |
| Intel® Core™ i9-13900 | Mainstream | 32 | 36 MB | 65W | Up to 5.6 GHz | Up to 4.2 GHz | 2.0 GHz | 1.5 GHz | 32 EU |
This is one of the problems of the situation, Intel should communicate the full list of affected models instead of giving generic indications that are subject to interpretation and can generate ambiguities.
I am completely out of my league in this conversation, but Intel's statement is consistent with what greymaterial suggested when he said C0 dies are likely not affected.
Facebook
Twitter
Instagram