Ugh, this has exploded on reddit and through a game of telephone seems to have turned a couple of anecdotes, a single photo of a damaged chip, speculation from users and media and a statement saying "We're aware of the reports and are investigating" into a massive issue that is "Confirmed" to affect all processors and spouting advice and recommendations based on it.
It's a great example of a hype train and something like citogenesis, each loop through comments and media seems to turn more speculation into fact.
It's not "a single photo of a damaged chip" there are at least 4 different public examples that I've seen. Not that it changes the scale of the evidence much.
The fact that the vendors are acting so quickly about this suggests to me that they have other non-public examples of the problem (from warranty returns, etc.) and the hullabaloo led them to compare notes and determine the root cause rather quickly. Or maybe they're just over-reacting to defuse the mob but I don't think that's super likely.
> The fact that the vendors are acting so quickly about this suggests to me that they have other non-public examples of the problem
In this day and age, if a story is spreading from a small thread on a subreddit to a string of reactions here and there in quick succession, a maker can't just let it run without any statement, even if it was literally a single guy with an unrelated problem.
Getting out a "We're looking into it" statement as soon as possible is the least they can do to manage the PR situation in any sane way, whatever the actual scale of the problem is.
I am only aware of 2 examples, the one that was posted on reddit and is now apparently being shipped to GamersNexus, and Der8auer, who didn't even notice pad discoloration (so may not have even been caused by an overcurrent event).
I wouldn't bet their "fast" response is due to them internally already knowing, so much as the PR hit from getting to the top of the PC enthusiast forums on reddit. None of the statements so far even suggest they've reproduced anything wrong, leaving vague statements of investigating it or adding more limitations in case that was an issue.
I remember a while ago something similar happened when someone read a lot into a comment into the open source AMD GPU driver as the "Shader Prefetch" feature of their GPUs was broken and disabled. AMD ended up having to issue a statement saying it's enabled and working where it benefits performance - and that shot to the top of those same forums and people "demanded" an explanation, and that was just an internal detail of hardware performance, not even a user-visible feature. To this day some commenters think the performance is somehow hobbled by this, so there may be more pressure to get faster responses.
I personally work on GPU drivers, AMD GPU drivers, but not on windows. I actively avoid commenting on things I have personal knowledge of, so I don't accidently leak something, but also because I see so much /wrong/ stuff repeated on the internet. In fact, I would say the majority of the specifics that are "generally accepted" on those subreddits are wrong to the point of being misleading. And a couple of times I've tried to question things I knew were incorrect I was heavily downvoted and dogpiled, so I gave up.
So while I have no direct knowledge of anything related to the CPU and it's power management, I'm trying to avoid Gell-Mann amnesia on "facts" repeated in those same subreddits.
So my current position is there may be an issue here, it may even be related to the SoC voltage regulation currently being blamed, but the "evidence" so far could also be caused by a thousand other things.
> To this day some commenters think the performance is somehow hobbled by this
There will always be a sizeable subset of people who are utterly convinced that there is an evil cabal of developers who refuse to press the "go faster" button(s) out of personal malice.
The four motherboard makers that have issued public press releases about revised firmware that is designed specifically to limit the SoC voltage to prevent failures would disagree with you here. If that isn't a public admission of this being the source of the failure, what is?
> The four motherboard makers that have issued public press releases about revised firmware that is designed specifically to limit the SoC voltage to prevent failures would disagree with you here.
What I'm seeing is four motherboard makers trying to win customers at the expense of ASUS by taking very public steps to state their product does not suffer from this issue unlike ASUS.
And I personally have an 8700k in my home machine.
I work in a large company, I'm not really involved with anything or anyone working on CPUs. I have no internal knowledge of anything in this area.
And I'm just pointing out how the statements of evidence presented here don't naturally lead to this conclusion as seem to be implied - not that this conclusion isn't correct, or some opinion about the quality of the hardware and software involved. I'm trying my best to not involve my opinion, and instead actually stick to the known facts. Which is my issue with this story as written.
I'm not sure if you're reading the same statements, but the ones I've seen explicitly don't admit fault.
> The fact that the vendors are acting so quickly about this suggests to me that they have other non-public examples of the problem (...)
That's quite the extrapolation. It's far more likely that AMD sees the exposure that a hand full of cases is having as a threat to the public's perception of the issues and is actively and very publicly addressing it to contain it.
So far all the reports I've seen about this issue was that it was really a problem with ASUS motherboards which were bricking the processor. The article seems to confirm it, and AMD seemed to have published a mitigation.
This is incorrect. Multiple instances (more than 6) have thus far been reported and pictured. All four major motherboard OEMs have now come forward with patches that are designed to limit the voltage. You can see their press releases in the link below.
These companies aren't making public statements about changes to their firmware to prevent failures for no reason.
Please don't do these kind of personal attacks. There are plenty of people in this thread with roughly similar opinions that "AMD employee kimixa" is expressing. Unless you want to claim they're all AMD employees it's a perfectly normal view to hold – if you disagree then you can say that without attacking people.
it's certainly not for no reason, but it could be a PR move not actually backed up by good technical evidence. The public evidence certainly most strongly hints that higher core voltages could be part of the issue, but the motherboard companies responding to that by doing the obvious thing does not necessarily mean they have actually confirmed the issue exists, just that they either think it's worth doing it just in case or as a way to avoid being seen as not doing anything about it.
This is a market segment which has had many a case of mass wild speculation and theories around a small case of reported failures, and most of the theories have turned out to be wrong.
We've got the same CPU (a nice little workhorse for a very gentle price I'd say)! I assembled the PC myself (hadn't done that in like 15 years for I had a shop do it for me but decided that this time, to not get too rusty, I'd do it myself) and put a Noctua NH-U12S cooler/fan on it, in a "Be Quiet!" tower (with a Be Quiet! PSU). So I like quiet fans.
Anyway I'm running Linux and put on all (but one) workspaces (or "virtual desktops"), the CPU governor in "powersave" mode, where apparently the CPU doesn't got above 3 Ghz and doesn't go above 45 C. Then on my "dev" workspace, the one where I need the performances, I set the CPU governor to "ondemand" and a single core can boost up to 5.51 Ghz (as reported by Linux) or all cores at once up to 4.9 to 5.1 Ghz with the temp reaching 95 C (which I think is how these 7000 series are meant to operate: they'll boost but then shall lower a clock a bit to not go above 95 C).
I configured my WM to automatically change the CPU governor depending on which workspace I'm on.
I'll never here the CPU cooler's fan when in "powersave" / "max 3.0 Ghz / max 45 C" mode. When the governor is in "ondemand" mode, I'll hear the fan indeed once I start compiling stuff.
I may, sadly, remove the RAM EXPO setting (seen that I bough DDR5-6000) while waiting for an ASUS BIOS / firmware upgrade for my Prime B650-Plus mobo.
It's not cool to run DDR5-6000 at 4800 but I'd rather not fry the CPU/mobo.
I considering a build with 13600k, and I am seriously just considering going with DDR5-4800. It's a tad cheaper than the DDR-5600 I was initially going to go with. I honestly do not think I would be able to tell the difference in day-to-day usage between DDR5-4800, DDR5-5200, DDR5-5600, and perhaps higher than that.
With that being said, I acknowledge that I am just making an assumption. Have you noticed any difference outside of benchmarks? I am not looking to go higher in speeds because I have seen videos (e.g. from Buildzoid) talking about how awful the 13th gen memory controller is, so I figured I am not really looking to gamble my time and money on max performance RAM.
Every time I'm setting up a new PC, I check the prices on RAM and then check the benchmarks for actual usage, and I always seem to find that buying faster RAM is not a good use of the budget, especially because I need a PC with a lot of RAM.
Currently running a 12900k with DDR4 RAM.
When I checked the benchmarks, DDR4 ended up being about 3% slower than the fastest DDR5 which was more than double the price, which for 64Gb was a significant price increase. Even less difference for gaming, something like 1%.
I'm sure it's a similar story for the different speeds of DDR5. The price premium isn't worth the tiny increase in speed. I guess people see the big numbers like 4800 vs 5600 and assume that means it's that much faster, 17%, when in reality it's something like 1-2%.
I'm really considering just going with a 12th gen, and upgrading in a few years vs. trying to stretch the 13th gen 3-5 years. It's a fool's errand trying to keep up with everything, and I need to do more research from a price vs. value point of view.
What motherboard are you using for your 12th gen? It's honestly been the most difficult part of piecing together my build. It seems like all the decent motherboards are plagued with one or more issues e.g. poor VRMs, poor bios/supporting software [1], NIC disconnects randomly [2], coil whine, etc..
I'll admit, I have only seen benchmarks for DDR5 in regards to games which was not impressive. It's disappoint to see that other applications would only have a greater performance of around 3%. Maybe I will just go with DDR4 then...
[1] I have heard Gigabyte boards have a hell of a time getting RBG Fusion 2.0 to work/save settings.
[2] I have also read to stay far away from boards with Intel's I225/226 NICs or to get a secondary NIC.
Gigabyte Z690 AORUS PRO DDR4. I'm very happy with it. Stable and does the job. BIOS seems good to me, no complaints.
Can't comment on RGB stuff because mine's in a closed black box.
Not sure what NIC it has... specs online say "Intel 2.5GbE LAN chip", my router is only 1Gb but it works fine, no disconnects. PC is Windows only though.
I've mostly had Gigabytes over the years (some Asus as well), and haven't had a lemon motherboard yet, maybe I've been lucky. But I am pretty demanding and don't tolerate faulty stuff well.
I'm also quite annoyed by noise, system is set up to be as quiet as possible with passive PSU/140mm fans/etc. Haven't noticed any coil whine.
I have not measured the impact outside benchs and yes you are probably fine with 4800s, but I use my pc for work and I would avoid slower ram when as a dev all my IDEs and processes do nothing but R/W from memory.
As a dev myself, do you think the difference between DDR4 and DDR5 is actually noticeable? That was mainly my justification for going DDR5 -- not gaming -- but I could step down to DDR4 if the the difference is insignificant.
> I have the same cooler! No expo issues on gigabyte ds3h.
Ah I'm less lucky: ASUS Prime B650-Plus mobo and the link says that on the Gamer Nexus forum (?) it could be a problem with ASUS motherboards.
> Why haven't you undervolted the cpu with all your interest for noise?
Honestly I'm not too sure how to do that. I never bother overclocking/underclocking. I like everything stock. I thought running DDR5-6000 from the QVL list in EXPO mode was also something really normal to do.
> Anyway I'm running Linux and put on all (but one) workspaces (or "virtual desktops"), the CPU governor in "powersave" mode, where apparently the CPU doesn't got above 3 Ghz and doesn't go above 45 C. Then on my "dev" workspace, the one where I need the performances, I set the CPU governor to "ondemand"
Is that a feature of a particular desktop or did you write a custom daemon to check on which desktop you are currently? I know that KDE has the concept of activities, which is an even different concept than virtual desktop and would be interesting in that regard.
>...how these 7000 series are meant to operate: they'll boost but then shall lower a clock a bit to not go above 95 C).
Bah, I have clearly been running my fans too hard. Using stock configuration, my 65W CPU idles in the low 40Cs with slightly audible fan noise. If I disable/super-slow RPM until 45/50 C, presumably that would be completely fine, and give me a near silent build. Obvious in retrospect.
Sorry about the OT but I would love to hear your experience with your build since I'm going to build a very similar one: AMD Ryzen 7k with DDR5 and the exact same mobo. My mail is in my profile
IMO, there needs to be more public shaming by game moderators when someone claims to have been banned for no reason and it turns out they were banned for spamming racial slurs or cheating.
Reminds me of the drama with NVIDIA's new power connectors burning out. Where IIRC it ultimately turned out to only involve a handful of cases where there was a manufacturing issue and the other reported cases were all loose connections by the users.
Thus indicating a design flaw of being prone to user error, but nowhere near the "fire hazard" type comments that were going around at the time (particularly on reddit).
> Thus indicating a design flaw of being prone to user error, but nowhere near the "fire hazard" type comments that were going around at the time (particularly on reddit).
wasn't that an actual fire hazard tho? Gamers Nexus just paid to proved the cause was because users werent connecting them correctly.
Yeah, it's a fire hazard when improperly installed, what I meant is that people were responding as if any nvidia card would just burst into flames even with everything properly connected.
Just to clarify: I'm not saying there isn't an issue, I don't know the "acceptable" failure rates and proportion of sales to judge, but there's certainly enough to investigate.
I'm just questioning the speed at which it went from "A couple of people had a similar looking failure" to "It's CONFIRMED to be due to bad SoC voltage regulation caused by enabling EXPO, and here's what you should do..."
Same with the NVidia filtering capacitor drama - there was an issue, but it ended up being fixed by zero changes to the "Accepted cause" of the filtering caps.
Or the NVidia 12HVPR cable drama - "Everyone Knew" it was due to vendors cheaping out on cables and the cable itself too close to the edge of tolerance to be acceptable for normal use. Until actual failures were analyzed and showed they weren't plugged in properly - which lead to a real fix (Future cables having shorter sense pins to ensure good contact of the power pins before allowing use).
>We are aware of a limited number of reports online claiming that excess voltage while overclocking may have damaged the motherboard socket and pin pads.
Well, if the overclocking part is true, shrug. Not that it's great, but it's also not completely unexpected for bad things to happen when you OC.
I personally don't like overclocking, yet I've always enabled EXPO, for many reasons.
First, because the producers of my RAM modules advertise it very clearly as feature; AMD advertises it quite loudly (https://www.amd.com/en/technologies/expo) as well. I think (not 100% sure) that it's even often set in RAM benchmarks .
Modern systems have very aggressive frequencies/dynamic frequency scaling, which blurs the limit of what constitutes overclocking. Intel sells unlocked models of CPUs that are supposedly can be "overclocked" (https://www.intel.co.uk/content/www/uk/en/gaming/overclockin...).
Many RAM producers sell modules based on their tolerance, so the fact that they're pushed to the limit - within the spec - is part of the sale.
Having said that, I definitely draw the line to using voltages (or anything else) that are outside the specification, and I consider crossing it overclocking.
Ultimately however, EXPO is a specification, so I don't consider it overclocking.
On AM4 many boards set too high IOD voltages when enabling XMP. Afaik no detrimental impact except about 10 W superfluous power draw.
It doesn’t seem particularly far-fetched that EXPO might suffer from the same issue except it accidentally sets batshit voltages instead of just too high. Afaik the SoC and IMC supplies are unsophisticated, there is no DVS going on, just BIOS setting a particular voltage in the VRM.
>Ultimately however, EXPO is a specification, so I don't consider it overclocking.
The O in EXPO stands for overclocking, and if that wasn't clear enough AMD themselves state using the whole word that EXPO is overclocking numerous times repeatedly.[1]
Indeed. Overclocking in 2023 is different than it was 20 years ago. You probably weren't overclocking back then without some inkling of the risks/dangers of eg. increasing the voltage to your CPU. And because you knew the risks you were probably stepping up very, very slowly and stress testing at each increment. Now you can kinda just grunt "me want faster computer" and push a button and it works. Unless it goes horribly wrong and catches your CPU on fire.
Chipmakers meanwhile can have their marketing cake by selling chips with certain properties, but pushing software that makes them "better than advertised" which seems like a great value. And then eat the legal cake by avoiding liability when chips fail because "running this hardware outside of the default specs is not supported."
And doctors recommend safe-injection sites, while also claiming that shooting heroin is unsafe. Shocking!
It’s called “harm reduction” — i.e. “if you’re going to do stupid thing X anyway, you may as well use our weak guardrails for it, to hopefully minimize the damage you’re causing compared to doing it through some random third-party solution.”
Oh please. They advertise their performance based on using things like EXPO. It's insanely anti-consumer for them to turn around and say "if you want to reach the performance we sold you this CPU on, you need to turn in this feature that voids your warranty and could potentially blow up your CPU". I don't understand why you're defending them here.
AMD has generally presented benchmarks with market-typical XMP/A-XMP/EXPO memory; their CPUs benefit more than the competition from that. Now some people have the urge to turn around and go “well why you’d do that, what are you, stupid?” for some reason.
> Well, if the overclocking part is true, shrug. Not that it's great, but it's also not completely unexpected for bad things to happen when you OC.
IIRC the core issue is that some motherboards loaded EXPO overclocking profile (basically, overclocking parameters for memory that are written on memory stick for "guranteed", "tested" OC performance) and caused overvoltage of CPU.
I don't remember if they found whether that's mobo reading profile wrong and burning up CPU or just some RAM sticks coming with too optimistic profile and mobo doing "well, I trust you, here are extra volts".
The wild part is that it is AMD tech (Untel has similiar equivalent in XMP), yet it also voids warranty as any other overclock
In the old days that would be true but nowadays OC is so mainstream that temperature sensors are expected to keep the SoC within safe limits, shutting down if needed but preventing damage.
Something escaped design models and test scenarios for this to happen with end users.
> In the old days that would be true but nowadays OC is so mainstream
I do not agree with this statement. If anything, overclocking is less prevalent than ever before, despite what all of the LTT's and friends make videos about.
The reality is modern CPU's are crazy fast, and auto-overclock (boost/burst) all on their own. There's little reason to push the CPU's further - and more significant gains can be had much safer by upgrading CPU generations than squeezing out a few extra percent of performance in benchmarks.
There will always be enthusiasts that want to overclock - but I don't think we can consider overclocking mainstream.
LTT has repeatedly said that overclocking under normal circumstances has been worth less and less as GPUs and CPUs have been increasingly built to ride thermal and power limits at stock. And that overclocking can even hurt performance by disabling that behavior.
For CPU clock, I agree the gains are marginal, yet BIOS settings still allow the end user to trivially further bump up the clock. There are young people with a lot of time on their hands who will do it for that 3% uplift.
Further, with memory we have things like XMP which again are blessed by the vendors and are 100% within the definition of overclocking. Just because a memory module can run at XMP rates doesn't mean the motherboard or the CPU have the signal integrity or power integrity margins to support the timings. And this is something component vendors can't optimize from the factory (only system vendors can), because it depends on multiple parts.
In the old days one couldn't enter BIOS setup and find even a single mention of overclocking, and the term itself was taboo because it meant operating the system outside the design spec. Nowadays a lot of vendors use OC in marketing. I think this alone makes my point about being mainstream.
In the article it said that apparently in some cases excess voltage damaged the temperature sensor which caused chip to not register overheating and just ask VRM for more power
Sounds like the thermal sensors don’t fail safe. If the thermal sensor isn’t returning readings or those readings are not within a reasonable range the CPU should shut down and then fail subsequent POST tests with a descriptive error.
I believe we've seen issues like this before from motherboard manufacturers, who are pushing more voltage to the CPU to get more performance out them. Recently is was on the Intel side of things.
It's not new- component OEMs just screw up sometimes.
I suspect that PGA sockets have less problem with this, though AMD has had just as long to get LGA right given their server processors have used it for nearly as long as Intel has so while it does happen, there really isn't any excuse for it.
I like to wait for the inevitable five minute king Linus Tech Tips video where Linus calls the guy at Gamers Nexus “really smart” and just repeats the findings that are somewhere in the hour long GN video. Because I only sorta care a little about this stuff.
There were few reports from non-asus ones IIRC, but it's hard to tell whether that was same "the EXPO profile loaded from memory fried it" or someone just going "MoRe VoLtS BeTtEr" and frying it.
Four motherboard makers have come forward and issued press releases about new firmware they are releasing that limits the SoC voltage, thus preventing this issue. What further evidence is needed?
The question is if they limit abnormal voltage, or just lock to "safe" ones.
this might not be a solution, this could be just a workaround basically crippling oc
AFAIK none of them said anything about what is happening nor what their fix is doing
> According to our sources and seconded by an ASUS statement to Der8auer, the problem stems from SoC voltages being altered to unsafe higher levels. This can be imposed from either the pre-programmed voltages used to support EXPO memory overclocking profiles or when a user manually adjusts the SoC voltages (a common practice to eke out a bit more memory overclocking headroom).
Does that mean, as long as I don't overclock, my Ryzen7 CPU will most likely be a-ok?
In general, you should be running the EXPO profile: this "upclocks" your RAM. I use the term "upclock" instead of overclock because you aren't pushing your RAM past its manufacturer's limit, you are pushing it to that limit. The JEDEC profile (the opposite of EXPO) is extremely conservative and can significantly affect performance.
This effect is compounded with Ryzen because of Infinity Fabric, which is the interconnect between the chiplets in the CPU. IF is clocked to the same speed as your RAM - so at the 6000MT/s "sweet spot" the article mentions your IF and RAM bus are running at 3000MHz. Matching IF to RAM bus is so important that increasing the RAM bus clock above the IF clock will decrease performance (unless the RAM bus clock is a multiple of the IF clock).
I haven't done the comparison on my 7950x, but on my 3950x running the IF at 1800MHz (RAM at 3600MT/s) resulted in a 10%-30% uplift in performance (depending on the workload).
So with Ryzen, if should consider purchasing RAM that is matched to the IF clock. For Ryzen 7 that is 6000MT/s. If you purchased RAM above that, you should definitely decrease the clock to match IF (you can tighten RAM timing to take advantage of the more expensive chips that you purchased, if you have the patience to).
What am I doing? I am going to disable EXPO, but I will definitely be re-enabling once this has been resolved.
Aside: What is MT/s vs MHz? DDR stands for "double data rate", meaning the MHz that everyone uses to describe the speed of RAM is a misnomer. "6000MHz" RAM actually runs at 3000MHz (but does two transfers per clock cycle). So: Mega-Transfers per Second.
>In general, you should be running the EXPO profile: this "upclocks" your RAM. I use the term "upclock" instead of overclock because you aren't pushing your RAM past its manufacturer's limit, you are pushing it to that limit. The JEDEC profile (the opposite of EXPO) is extremely conservative and can significantly affect performance.
Any source I could find says that EXPO does void warranty, as it is overclocking. https://www.amd.com/en/technologies/expo (in footnotes section at the end of page)
It might move RAM to it's manufacturer spec as you said but it does it by overclocking CPU
> So with Ryzen, if should consider purchasing RAM that is matched to the IF clock. For Ryzen 7 that is 6000MT/s. If you purchased RAM above that, you should definitely decrease the clock to match IF (you can tighten RAM timing to take advantage of the more expensive chips that you purchased, if you have the patience to).
Do you know any good benchmark (preferably Linux one) that would show those differences? I will be getting 7800X3D in few days so I want to test it for a bit.
> What am I doing? I am going to disable EXPO, but I will definitely be re-enabling once this has been resolved.
The core of the issue seems to be "just" overvolting so I'm guessing any manual overclock would still be fine. I have no need for it now (just puny 1070 to pair it with for now), but yeah, wait for new BIOS firmware seems to be best option.
> Aside: What is MT/s vs MHz? DDR stands for "double data rate", meaning the MHz that everyone uses to describe the speed of RAM is a misnomer. "6000MHz" RAM actually runs at 3000MHz (but does two transfers per clock cycle). So: Mega-Transfers per Second.
Not exactly. The specified speed is only speed of the bus between memory and CPU, not the speed of memory chip itself. DDR4 3200 and DDR5 6400 have exact same internal speed (400MHz) and this is why there isn't much of latency improvement between generations (DDR5 does have some other tricks for latency).
Also the 6400MHz is actual speed data is put on the bus, even if clock is half of that.
It's pretty much entirely so the RAM chips don't have some insane number of pins but something manageable.
> Any source I could find says that EXPO does void warranty, as it is overclocking. https://www.amd.com/en/technologies/expo (in footnotes section at the end of page)
> Overclocking and/or undervolting AMD processors and memory, including without limitation, altering clock frequencies / multipliers or memory timing / voltage, to operate outside of AMD’s published specifications will void any applicable AMD product warranty, even when enabled via AMD hardware and/or software. This may also void warranties offered by the system manufacturer or retailer. Users assume all risks and liabilities that may arise out of overclocking and/or undervolting AMD processors, including, without limitation, failure of or damage to hardware, reduced system performance and/or data loss, corruption or vulnerability. GD-106
I am not a lawyer! BUT!
AMD voiding the warranty like that sounds extremely illegal.(Depends on your jurisdiction, obviously.) EXPO is very clearly advertised on the box, documentation, and media coverage. It even comes on by default in a number of cases. That would mean the AMD Processor/Memory would have the warranty voided when you first power it on!
The EU specifically has a law covering that any advertised features are covered by a legally mandated 2 year warranty.
Well, it is an advertised feature that explicitly states (althought under hidden footnote) that it breaks warranty so yeah, would be nice if someone tried to get them to court over it. I'd imagine they would be raked thru coals in EU
> It even comes on by default in a number of cases. That would mean the AMD Processor/Memory would have the warranty voided when you first power it on!
That I'd be doubtful about as AMD have no bearing about what options random motherboard vendor enables by default
I was still using Windows when I did everything with my 3950x. I used AIDA64 for the memory, specifically paying attention to latency (which is the strongest signal you will see for IF 1:1). Other than that it was a few tests that matched my anticipated workload: pi for stability, 3DMark for games (another decent signal), compilation speed for dev.
On Linux+7950x I just did some subjective tests (I have less time these days) with rust-analyzer responsiveness and there definitely was an improvement, but I can't quantify how much.
I might be less of an issue for a 7800 because, as far as I remember, it doesn't use chiplets.
IF maxes around ~2GHz on current chips so it is impossible to run it 1:1 on anything faster than DDR5-4000, but apparently something was improved along the way, and optimal RAM speed is at mentioned 6000 and not matching doesn't hurt as much in Zen 4
it appears that even with mismatched clock (the 2166 IB with 6000MHz memory) latency continues to drop, and problems start if you try to run memory controller above that
My reading of the statement (and understanding of other reporting on this subject) is that EXPO alone can trigger this issue. Calling EXPO overclocking feels somewhat disingenuous to me.
EXPO is akin to Intel's XMP. The RAM stick reports timings and other settings that it can handle and then the motherboard/user selects one to use. The profiles are needed because the official JEDEC standard for DDR doesn't provide a mechanism for running up to the frequencies that modern RAM uses; e.g. if you buy a DDR5 6000MHz kit, it'll only run at 4800MHz or thereabouts until you enable EXPO/XMP. RAM is really never expected to be run at the base frequency (without EXPO/XMP). If you buy a prebuilt, it'll have EXPO/XMP enabled, and if you build yourself you always should be enabling it. If you have a RAM kit that's somewhat recent and don't have it enabled, you've likely wasted money on the RAM.
Intel has attempted to claim that XMP was overclocking and violated their warranty but my understanding is that they quickly backed off.
> My reading of the statement (and understanding of other reporting on this subject) is that EXPO alone can trigger this issue. Calling EXPO overclocking feels somewhat disingenuous to me.
Totally. With DDR5-6000, it's even nastier: AMD repeatedly said that DDR5-6000 was the "sweet spot". I never overclocked any computer of mine but I did buy DDR5-6000 for my 7700X because AMD said it was the sweet spot. And I turned EXPO on so that it'd run at 6000, not 4800.
And now, after saying 6000 was the "sweet spot", they try to word things as if EXPO was somehow overclocking!?
I'm in same boat. I payed a premium for DDR5-6000 64GB for my 7950X, because AMD said it's the sweet spot... I didn't ever consider this "overclocking", I haven't ever been interested on overclocking, I just want stable machine.
Now I'm considering disabling EXPO, but I don't think it would work as 6000 anymore after that. I have used this machine for 4 months now, so I'd say it's pretty safe, but who knows, I haven't taxed the CPU with gaming yet.
> Overclocking and/or undervolting AMD processors and memory, including without limitation, altering clock frequencies / multipliers or memory timing / voltage, to operate outside of AMD’s published specifications will void any applicable AMD product warranty, even when enabled via AMD hardware and/or software. This may also void warranties offered by the system manufacturer or retailer. Users assume all risks and liabilities that may arise out of overclocking and/or undervolting AMD processors, including, without limitation, failure of or damage to hardware, reduced system performance and/or data loss, corruption or vulnerability.
On their own page they consider it warranty-voiding feature that they ADVERTISE and PROMOTE
> Now I'm considering disabling EXPO, but I don't think it would work as 6000 anymore after that. I have used this machine for 4 months now, so I'd say it's pretty safe, but who knows, I haven't taxed the CPU with gaming yet.
I'd drop it, wait till it all blow over, upgrade the BIOS, and re-enable it. It does seem like just a firmware bug or maybe some OC profile on memory being overly optimistic with voltage.
That's not how it is presented in the BIOS. EXPO settings aren't under overclocking, to me they didn't appear scary overclocking thing.
In BIOS there is an separate section "Overclocking" and to get there you have to "Accept" or "Decline" some big text about warranty, I never went there to enable EXPO. So I never needed to press "Accept" to void my warranty to toggle on EXPO...
EXPO settings just to my eye looks like this:
EXPO II turned on, memory frequency 6000 MHz (what is says on the memory), and CPU frequence stays where it was (thus not overclocked).
> EXPO II turned on, memory frequency 6000 MHz (what is says on the memory), and CPU frequence stays where it was (thus not overclocked).
It's a technology to overclock memory controller on CPU, not CPU cores themselves. Increase of voltage of memory also needs to be applied to the controller, hence where the problem in article comes from.
But yeah, I can see someone arguing that in court and winning, with AMD only warning about warranty loss in footnotes and mobo vendors not presenting it as such either.
This kind of corporate behavior is common. Subaru for a while offered free SCCA memberships with purchase of a WRX so you could autocross it, then you do an autocross and they try to get out of any warranty work because you autocrossed it.
Certainly tire/brake pad wear shouldn't be covered but they were more ridiculous about it than that.
Be sure you know what autocross is before you type up a big lecture about wear and tear from racing. Also I already know.
My Chevy came with instructions on proper track prep and tracking it (if prepped as instructed) won't void the warranty. Didn't include free track days though.
Autocrossing is more my thing and I'm shocked any car manufacturer would try to weasel out of warranty for it. I've seen modified cars break and even catch on fire, and can understand racing tires could put undue stress on stock suspension, but a stock car that can't handle autocross? Should we expect parts to be falling off if you need to do an emergency brake or evasive maneuver on the road?
>if you build yourself you always should be enabling it.
Absolutely not. This is such a terrible advice I would think you were trolling if I didn't assume better of you.
An end-user should not engage in any overclocking, XMP/EXPO included, without thorough knowledge and understanding of what they're getting into. You are literally driving your hardware above and beyond official, published specifications.
And no, marketing is not "official, published specifications".
> You are literally driving your hardware above and beyond official, published specifications.
That is not the case if we talk RAM. It is sold as 6000MHz, it is tested as 6000MHz and it is validated to run at 6000MHz. And there is a standard each developed by both Intel and AMD to allow RAM to ran at it's rated speed.
It isn't just marketing. The problem is, that this standard entails going out of spec on the CPU side, where it does mean going beyond validated limits. But the consumer shouldn't have to dig to understand, that these standard designed to make your RAM run at it's rated speed technically means pushing limits.
I think this is a bit of a stretch from tomshardware, as if you actually read the statement https://i.imgur.com/pM358j6.jpg it doesn't actually say what the problem is, or even that there is a problem. Just they released a new BIOS that has more limitations.
And the other "evidence" of a photo of a damaged pin pad on a ryzen processor shows the damaged pins aren't related to the SoC rail mentioned in the article.
The only thing I can see in the article that supports that is:
"Our sources also added further details about the nature of the chip failures — in some cases, excessive SoC voltages destroy the chips' thermal sensors and thermal protection mechanisms, completely disabling its only means of detecting and protecting itself from overheating"
And "Our Sources" aren't motherboard vendor statements, none of the statements from motherboard vendors or AMD themselves support this theory, so we're relying on Tom's unnamed "sources" rather than any public evidence or statement.
This is exactly what I dislike - the slow removal of the source and quality of evidence and information. It's Toms being bad too, as their statement
"seconded by an ASUS statement to Der8auer, the problem stems from SoC voltages"
does not seem a good representation of the actual statement - the image of which I linked.
ASUS pointing out that they are defining new rules for vCore and SoC voltage to address these issues is in fact pointing to that as the issue. They aren't just "throwing that in there" because it is unrelated. If you need it to be even more direct, here is the statement from ASUS that says it is SoC voltage.
You've admitted that you work for AMD, which would make a casual observer think that you are being biased in your comments. There is verifiable proof of the correlation with SoC voltage from the ODMs themselves.
Vendors are putting out updated BIOS that limit voltage going to the chips. Based on internet hearsay, even some of the non-OC chips were getting wrecked by stock BIOS overvolting enough. The real concern is with the X3D chips which are particularly sensitive to voltage and not particularly meant to be overclocked.
I think you're fine as long as you flash an updated BIOS put out by your MB manufacturer.
I guess that depends on the nature of the failure and the fragility of the part.
If you're staying inside the recommended voltage its likely the product will have a long life. For example look at the life curve of capacitors (and I'm not talking the diseased ones we had to deal with in the 2000s). The 'hotter' we run them the shorter their lifespan is. Now what can be difficult to figure out via testing is to figure out the MTBF when you have a hard cutoff point for failure that's not very much higher than standard operating specs. We could be failing at 100 hours at 1.4v, and 10,000 hours at 1.3v. Or it could last 100,000 hours at 1.3v.
What I'm not sure of is DDR5 - it had seemed DDR5 offered more flexibility in voltage/speeds, i.e. the "stock" ddr5 speeds at 1.2V seemed conservative compared to 1.35 or 1.5V for DDR3 or DDR4. But if going about 1.2V or whatever voltage DDR5-6000 EXPO profiles are setting, also necessitates increasing SOC/CPU voltage...maybe it's better to hang back and just stick w/ DDR5-5200 or 5400 or something like that.
That’s the root of the problem seemingly - going past the official spec (5200 MT/s for 2 sticks, 3600 MT/s for 4 sticks) requires additional voltage to the memory controller and it’s not uncommon for motherboards to automatically punch up the voltages when you turn on XMP or Expo. You requested the board to overclock after all, they’re just helping you get it stable without tinkering! It’s completely legal from the mobo vendor’s perspective to enable more voltage in “auto-OC” scenarios and they are in fact incentivized to do so by the possibility of returns and negative reviews/bad word of mouth. Customers will remember when “it didn’t work on Asus but I just enabled XMP on MSI and it worked fine”, even if that’s because MSI is punching up voltages. They will see that MSI scores 3% better in whatever benchmark. And AMD themselves in fact advertises benchmarks with Expo enabled, despite the fact that it officially voids the warranty. AMD and Intel (both) have gotten away with this for a long time, having their cake on heavily implying/suggesting it be enabled but technically voiding the warranty if you do (although it's quite unenforced unless you openly tell a warranty agent that you did it).
I had a 9900K fail due to simply enabling XMP too, and I’ve seen many reports of early Zen1/Zen+/Zen2 with what seems likely to be IMC failures (they are getting to be "of the age" - 3-5 years with increased VSOC and the memory controllers are just clapped out). DDR4 controllers seem more delicate in general compared to DDR3, and DDR5 requires still higher voltages. It’s almost shocking to hear buildzoid recommend 1.3v VSOC as a “safe” daily driver. AM4 voltages were more like 0.8-1.1v.
It also doesn’t help that the early AM5 IO die seems to be terrible. AMD's memory controllers always been worse than Intel even on AM4, but a lot of AM5 chips won’t even post with 4 sticks even at the pedestrian official spec of 3600 MT/s (intel is 3600 MT/s too but they actually work mostly reliably at it, it's a lottery with Zen4). Zen5 is supposed to feature an improved IO die and I wonder if the VSOC will be lower there too.
I'll leave my little anecdata with Intel Alder Lake memory controller shenanigans since it's relevant.
I built an i7 12700K system last year with 4x16GB DIMMs of DDR5-5600 RAM, for a total of 64GB RAM. Not being fully aware of the memory controller's limitations at that point (moral: read the datasheets!) I turned on XMP which obviously overclocked the RAM to 5600MHz. Note that at this point I was curious why the mobo had booted the system with RAM clocked to 4000MHz prior to my fiddling around with the BIOS.
Afterwards, I had instability which manifested in random program and system crashes. Eventually I checked the RAM and confirmed that was the issue with memtest86 returning errors. I read up on the CPU's datasheet and learned that when supplied with 2 DIMMs per channel with 1 rank per DIMM, the memory controller downclocks to 4000MHz as specified. Incidentally, this answered why the mobo booted the system up with RAM clocked to 4000MHz at the beginning.
I turned off XMP and backed off the overclock to 4800MHz with timings as applied by the mobo (they're just JEDEC profile timings), and that got me back to stability with subsequent tests in memtest86 returning no errors.
The important things to takeaway here are that:
* DDR5 memory controllers are still in their infancy compared to DDR4 and DDR3 controllers today, and can't yet deal with so much RAM at once.
* Going beyond 1 DIMM per channel, 1 rank per DIMM means the memory controller will run slower than advertised, and this is according to specification. It is on us the end-users to do our homework, we can't forget we are using precision electronics.
* With the memory controller running slower, overclocks are that much more taxing than XMP/EXPO and the marketing would imply. Simply turning on XMP/EXPO without thought is dangerous because those profiles do not account for a slower memory controller.
* Overclocking is by its very definition running hardware above and beyond what they were designed and rated to operate at. Warranties will be voided, and hardware can be damaged or destroyed because they are operated out of specification. You have been warned, ignorance is not an excuse.
* Between CPUs (and GPUs) today coming "over"clocked from the factory to perform at their peak and RAM speed having negligible real world performance implications for the vast majority of use cases, end-user overclocking is simply not worth the hassle unless you absolutely know what you're doing.
* EFI/BIOS firmware nearly always come out of the box with safe defaults. If you don't know what you're doing, leaving them alone is perfectly fine. You will still have a very performant system. Overclocking is not for the faint of heart (nor the ignorant!).
"Probably", but not guaranteed. This isn't a normal thing to have happen, even with overclocked chips. The temperature sensors used for thermal protection aren't usually the first thing to break.
It might not be temperature sensor itself but circuitry measuring it. Yeah sensor is "just a diode" but that needs amplifier and ADC to get to the rest of the system
Old days like 1st-gen Ryzen in 2017? The launch chips were defective. I confirmed it with my own and had to get a new one.
> AMD has not provided an official public explanation of the fundamental problem, but [...] it appears to affect Ryzen CPUs manufactured prior to week 25.
This looks like "we didn't told anyone about limits they should not cross" or maybe had too optimistic ones in NDA'd docs they passed to the mobo vendors.
Or mobo vendors dropped the balls and thought those were "recommended" and not "absolute max"
Early Zen2 chips were also basically defective, 3950X and 3900X advertised a very high boost clock that was seemingly impossible to reach in practice. Some chips did, some didn't, and seemingly it came down to silicon lottery whether your chip would miss advertised clocks by as much as 10% like GN's did.
There was a series of patches that didn't really resolve anything or change performance in any notable way, and people proclaimed it not actually a problem in the first place, but definitely fixed after this series of patches to fix the not-a-problem. Later batches did seem to have better silicon quality and everyone kinda forgot about it and moved on, because everyone constantly gives AMD hall passes on this kinda stuff, but the people who had bad chips didn't have their issues resolved, and the same boards ran fine with later chips.
I mean basically they sold you a 3900 non-X for 3900X prices and everyone searched for reasons to give them a pass: "it says up to 4.7 GHz", "it runs 4.7 GHz if I run a loop of NOPs", etc. And this was very unprecedented at the time, we hadn't normalized the culture of boost being this variable thing, every single Zen1/Zen+ and Intel chip would do the advertised clock rates at least.
And the whole thing could have been avoided if they had just advertised a more conservative clock and let the people with good silicon be happily surprised that they were hitting 10% over box speeds. Very anti-consumer then too. As GN themselves said in the first link - this surely would have been a massive scandal if Intel had done it but everyone gives AMD passes constantly.
This recently resurfaced a bit with Zen4 again. Silicon quality matters a lot again, der8auer has found that perf/w can vary by as much as 50% (!!!) based on silicon quality. There is a range of performance (~10% let's say) and then bad chips may pull as much as 30% or so more power than a good sample even to get into the bottom of that range, while a great sample will trounce their way to the top of the range while sipping power. Again I would expect this to decline over time as the silicon matures, and there isn't really any direct advertising or promises about perf/w, but... silicon quality is very much in play once again, especially in the lower bins.
Anyway though the Zen1 segfault bug you mentioned was brushed under the rug too, once again. There was nothing Linux-specific about it, compiling in particular seemed to trigger the bug more easily (this is unsurprising: compiling is very cache-intensive and uop-intensive and often shakes out silicon bugs of all kinds!) and people would have random problems/crashes/etc when compiling under windows too. The processors were simply defective (AMD quietly made changes to the uop cache iirc) but they worked well enough that most people didn't notice. Hey, if you randomly crash out of your game once a week, how are you ever going to tie it back to silicon defects? It's a Linux problem, doesn't affect me. Or those crashy RDNA1 drivers that "worked fine for some people" and after much to-do and a bunch of supposed fixes still never quite worked right for some people... :thinking:
(which of course AMD got a pass on the 5700XT too from most people! like, this ryzen 7000 problem is just a bios bug most likely, but they do really ship openly defective products quite often and don't necessarily always fix them fully or make people whole, and they usually get away with it and people defend them while they do it. The criticism is what gets fixes and keeps them honest, they're a company and if they keep getting away with it they'll keep pushing, just like people criticizing NVIDIA for VRAM size or whatever is ultimately what leads to improvements on that in the future. It drives me insane when the AMD crowd just blindly defends and justifies anti-consumer behavior because that makes it harder to get actual improvements!)
Anyway, the rest of this is kind of bitter and jaded, but, the AMD fan club has been kinda awful since forever. It's leftover bunker mentality and chip-on-their-shoulder from the Bulldozer days and early Ryzen days, but like, Steve from GN and der8auer and LTT and others have commented on it repeatedly over the years... like the AMD defense force justifying AMD cutting off all legacy chipsets (B550 had not even launched yet and they wanted to kill X470 and B450 support for Zen3, despite B450 being the "current gen" budget boards!) which triggered this video. And they got away with killing off X399 and doing a single-gen TRX40 socket (despite AMD's promises of it being a long-life platform to justify breaking compatibility) and just abandoning HEDT in general, and doing huge price increases during the pandemic (while Intel cut prices). They doubled ASPs of their 6C within a matter of a few months, they cranked the shit out of it and profiteered during a public health crisis while Intel held prices and in fact cut a lot of them too.
And sometimes they do make people whole (segfault bug you could send in your chip if you cared), but like, other times they don't, 5700XT never really got fixed, Zen2 launch silicon never got fixed, etc. And then people shrug and move on, and keep defending them as inherently and consistently pro-consumer.
Paying studios not to implement (and in some cases rip out existing) DLSS2/3 implementations is another bullshit move lately too. They are betting if they just lean on their console marketshare and stonewall DLSS as much as possible, people will eventually just come to accept a worse outcome, because it's easier for studios to not have to validate two things/etc. Even if the API is open they're not going to accept anything that could be used to plug non-free code, even going as far as pushing for statically compiling FSR2 into everything to prevent DLSS Swapper/DLSS wrapper integration by swapping DLLs around. Anything that could be used to plug something they don't like is bad for them, so it's bad for you.
Or the Ryzen take-a-way and PREFETCH speculative execution errors... which have vectors which remain essentially unpatched (some were, others weren't) apart from enabling KPTI, which AMD continues to recommend be disabled by default on their architectures. So they're shipping an insecure configuration and "if you care, turn it on". Imagine if Intel did that with Meltdown, shipped an insecure configuration "to win benchmarks"? People would lose their shit.
Like business is business, profit-seekers gonna profit-seek, but people defend their anti-consumer behavior constantly and it ends up hurting consumers because it avoids AMD having to face criticism and actually reform their ways. They'll keep doing it as long as they keep getting passes.
(and specifically der8auer low-key calls out the same integration between AMD marketing with social media later in the video. Intel and NVIDIA don't have a "red team plus", they don't have a "pathfinder program", by application, for their most stalwart social-media defenders. https://youtu.be/x03FyPQ3a3E?t=356 )
AMD really also kinda normalized the idea of marketing around base-TDP... this existed for laptops really was not a thing with desktop processors, 5960X and 7700K etc pretty much stayed within TDP while boosting even in prime95 (which shouldn't be surprising given the clocks/core counts! 5960X is like a 3.3 GHz all-core boost, and 7700K gets 91W to boost 4 cores with an AVX offset as well!), then AMD redefined the term TDP and introduced PPT as their "boost TDP" and then spewed a bunch of crap about how thermal watts are not electrical watts (thanks AMD_Robert!) and thermodynamics do not apply to AMD. Anyway, all they had to say was that PPT was their boost TDP and the chips would run above the official TDP as thermals permitted, but they wanted to stay far away from that "TDP" term to build themselves a 30% cheater factor on all their power measurements. To this day people say things like "I have a nice efficient 65W processor and Intel pulls 100W" when the difference is really more like 90W vs 100W or whatever, it absolutely worked and people continue to buy and mentally follow the marketing number instead of actual power.
Not that they're worse than anyone else. Intel and NVIDIA say and do some shit too. But people think AMD's shit don't smell and they've kinda done some anti-consumer crap over the years too and the fan club kinda drives me crazy defending it. It's nuts how much people love AMD and will defend pretty much anything they do. If I sound jaded it's because the last 6+ years have been nonstop Narcissists Prayer from the AMD fan club: "that didn't happen, and if it did it wasn't bad. And if it was, it's not a big deal. And if it is... it's not AMD's fault. And they didn't mean it. And if they did... you deserved it."
Even stuff that is pretty clearly AMD like the chipset lockout or brand-locking OEM chips to kill the secondhand market (how does brand-locking prevent swapouts? ok I have to find an asus branded 200GE to swap in when I steal this 5950X, so what?) gets defended with "but partners must have demanded it to sell more motherboards, AMD is just a small family company and can't stand up to the likes ASUS and MSI and Gigabyte!", when the track record pretty clearly supports partners being willing to go much farther on compatibility (like enabling PCIe 4 on some B450 boards that have good enough signal integrity) and AMD being the one putting the foot down to stop them for marketing segmentation.
Or the "the reason AMD can't offer good-value GPU products is because you didn't buy them in the past, they're just matching NVIDIA's pricing because they know they can't take marketshare! you deserve it for not instantly giving AMD 80% marketshare that one time they were ahead for 6 months!", etc. High prices aren't caused by AMD being an amoral profit-seeking entity like every other corporation... it's your fault because the 4850 didn't give them overnight market dominance, they want to do great prices but know that gamers won't buy it!
AMD literally can do no wrong that social media won't minimize, deflect, or shift blame to someone else. It is straight-up Narcissist's Prayer shit constantly. Drives me nuts. It's fine to like AMD, or even think they've been a net positive for the market (they have!) but like, you know that I need to throw in some praise at the end because the fans are unhinged, and that's the whole problem, you can't even complain about outright instances of anti-consumer behavior without polishing AMD's knob afterwards, or someone throws a hissy because I besmirched AMD's e-honor.
And that's crazy, nobody would ever do that for most other tech companies - imagine if criticism of GPP had to be accompanied by similar "kissing the ring" with platitudes of how much NVIDIA had done for advancing graphics tech over the decades, or criticism of Smeltdown had to be accompanied by buttering up Intel. Social media's relationship with AMD is incredibly unhealthy. And people like GN Steve and der8auer have called it out as such too, and it just keeps rolling on.
If Intel did something like brand-locking chips there would be nobody defending it, or saying that OEMs want it, or that it's a security feature (again, how does requiring me to find a Lenovo branded chip prevent swapouts?). Daisychained chipsets, on boards that clearly don't justify it? Some of which literally have a single M.2 slot on the whole second chipset, that could have just been connected directly to the first chipset? Sounds like AMD wants to sell more chipsets. Locking out legacy chipsets for bullshit technical reasons then quietly retracting the whole deal later? Yeah, probably sounds like they wanted to sell more chipsets. But like, you have to meticulously eliminate any possible defense or beneficial motive, how dare you impute a profit motive to any of AMD's actions. Again, just drives me nuts. If it were Intel we’d just call a spade a spade and say brand-locking is pretty obviously being pushed with killing the secondhand market as a strong secondary motive if nothing else. In fact people said exactly that when intel locked xeons out on the consumer chipsets (after Broadwell). No defense or knob-slobbing required - they did it for the money.
Again none of this is really about the Ryzen 7000 bug. I'm sure it will be patched. People with dead CPUs will get them warrantied. This isn't a showstopper. It's just a general trend in the way these crises get handled for AMD vs everyone else, the Narcissist's Prayer routine gets old.
Woah, you have the receipts! My PC hardware enthusiast phase is behind me, but I definitely agree with you that there was/is a weird sickly "AMD defence force" out there. I think it's just another manifestation of people strongly identifying with their pro sports team.
> Our sources also added further details about the nature of the chip failures — in some cases, excessive SoC voltages destroy the chips' thermal sensors and thermal protection mechanisms, completely disabling its only means of detecting and protecting itself from overheating. As a result, the chip continues to operate without knowing its temperature or tripping the thermal protections.
Spending a month desperately trying to kill a CPU that is known to be extremely sensitive to over-volting finally nets a result. Congrats to the Intel shill Der8auer.
I am glad there are armies of tinkerers but this guy knows damn well these CPUs should not be overclocked. He was trying to kill it
Why shouldn’t they be over locked? The chip makers have made it easier by the year to the point where AMD releases multiple windows programs to their users that provide newbie friendly overclocking. Adrenalin and Ryzen Master. Not to mention that basically ALL 3rd party GPUs are factory overclocked
Context is 3D cache CPUs and they shouldn't be overlocked. 5800x3d cannot be clocked because options in BIOS/UEFI are disabled. Ought to be the same with AM5 but this guy found a way to do it. As expected, the CPU died which, as luck would have it is exactly what AMD said would happen
The press releases from the motherboard vendors are 100% proof that these failures are real and a problem. They all cite new firmware that limits SoC voltage, thus showing the analysis in the article is correct.
the new BIOS limits SOC voltate at 1.3V, but the whole point of this "DDR5-6000 sweet spot" claim is the ability to run at 1:1 with the memory controller/memory frequency, i.e. 3000 mHz (from DDR5-6000 Mt/s) to 3000mHz UCLK memory control. Infinity fabric, it appears, runs at its own frequency, 2000-ish mHz or something like that and is not tied to above.
So: with the BIOS limit to 1.3V SOC, will the max memory controller frequency be less than 3000 mHz? Is the new sweet spot then something like DDR5-5600?
It's a great example of a hype train and something like citogenesis, each loop through comments and media seems to turn more speculation into fact.