Intel has tacitly approved up until now. Since there is now a hue and a cry they are changing their tune. But lets be real they knew and approved or they would have done something about it a long time ago. Its only the negative press (And possibly Potential RMA concerns ) that cause them to change their tune.Reply
Here in Chicago my solution was to run my 13900k (which was never stable enough to launch Battlefield 2042 and would occasionally greet me with a BSOD otherwise using stock Asus XMPI profile) was a mild vcore offset of +0.050. This obviously increased temperatures in the Winter and wasn't a problem until this last month when it became difficult to keep my room under 80F with my HVAC system while gaming. The RTX 4080 doesn't help, but I measured the CPU produces an extra 100w of heat when using my offset with XMPI (Asus profile) verses stock offset and XMPII (Intel profile) and negligible performance difference.
I don't plan to bother running XMPI anymore, period. Whatever performance advantage the Asus profile offered was indistinguishable from Intel's parameters and not worth the hassle.Reply
I have thought the same thing. Intel knew and supported unlimited power to CPUs just so they could win benchmarks. However, that has finally caught up to them as 13th and 14th Gen P-Cores are all versions of 12th Gen Golden Cove just pushed to the absolute limit. This what they did with Skylake as well and also why Intel cannot go with 16 P-Cores as power consumption would be astronomical at the clock speeds they need to be competitive. Reply
You don't know that. You don't actually know anything, you just like to make random statements without bothering to care if they are true or not.
Intel has always had more rigorous product validation than any other chip maker. That's a fact. The fact that there are bugs is just the nature of complex machines.Reply
It's nothing new with them. The Pentium III that broke the GHz barrier, after the Athlon, was unstable and had to be recalled. And early Northwood steppings would degrade when overclocked and at length fail.Reply
We haven't decided quite how we'll go about things yet, since that's going to hinge in part on what Intel's official determination in May ends up being - and whether mobo vendors actually change their out-of-the-box defaults or not. But yes, at some point we'll have to pull in some additional benchmark data with the baseline settings.Reply
You really should re-benchmark. It would be consistent with a statement you regularly made about testing CPUs using JEDEC profiles instead of the XMP ones, because XMP was considered as some sort of out-of-spec overclocking. So shouldn't a similar policy also be applied to other system components, like the CPU?Reply
As things stand, per Intel's warranty terms, the aggressive settings by motherboards are not overclocking the CPU. XMP, on the other hand, explicitly does violate the warranty.
Overall, our preference is to test out-of-the-box settings, with the caveat that we won't void warranties in stock testing. So aggressive power settings, which are covered, pass that criteria - though I certainly won't mind if Intel reins in the default settings. Conversely, mobos don't enable XMP by default to begin with, so enabling it would also violate the OOB principle (and even then, I have to admit we'd probably still enable it if Intel's warranty covered it).
Obviously, all of our current settings go out the window if Intel changes the OOB experience.Reply
Just in case it needs pointing out for the millionth time, this site's policy of not enabling XMP in benchmarks, makes the benchmarks useless to its audience.Reply
“Useless” is maybe a bit strong - mainly because the benchmarks on the site at least all have an internal consistency amongst themselves (which I’m sure factors heavily in their decision to uphold it), but I agree that it seems a bit antiquated and limiting today.
They’ve never claimed papal infallibility, so they could change their mind at some point. In the DDR4 era, it seems 3200Mhz CL16 or 18 would have been the logical choice. Maybe a bit hard to say right now for DDR5, which again probably factors into things.
But given they seem to run this site on a shoestring, I’m kindof just glad they are still in business and there’s other things I’d rather see… like less ad-laden slow-rendering pages and clickbaity-style multi-page delivery. Heck, even in trying to write this comment I had to start over because the safari tab crashed and reloaded the page.Reply
What would interest me more is a guide to how we can determine the proper default values and how/where to set them ? F.e. ark.intel does not specify much information on the CPU page. So HOW can I validate that the MOBO maker actually implemented the baseline profile correctly ?Reply
Intel is just throwing their unstable 13th and 14th gen silicon under the rug of OEM BIOS options.
Historically speaking as AT already noted Intel PL1 and PL2 and the TAU were there ever since Core series Architecture went mainstay, like since Sandybridge onwards from what I can recall. And Intel Power Limit was never stock and none of them failed like these.
Now looking at Buildzoid videos and other forum talk (OCN, Notebooktalk), Intel pushed these processors to extreme for their marketing team so that they can upsell these despite higher heat density and power consumption which is far higher than Intel 11th gen Rocketlake 14nm backport. 13th gen is impossible to cool on an AIO, whereas 11th can barely manage, 10th can do it. And 13th / 14th gen are extremely hot on even an AIO, the reasons are - Intel 6GHz marketing, their new Ring OC pushed beyond 5GHz (10th gen can OC past 5GHz but they had infamous WHEA issues, only way to fix was either get lucky by silicon lottery top bin of SP rating or you degrade the cache) and XMP.
Now the 13th / 14th gen processors binning is atrocious, for this reason many processors cannot finish Cinebench R15-R23. Due to the silicon instability out of the box, generally speaking most of the silicon is given higher voltage to cope up with the mean distribution of avg voltage across the CPU arch, and still that is not enough. Until now Intel Core processors never had such instability issues OOTB. Stock or MCE, they never exhibited this extreme instability, and matter of fact Intel CEP, Current Excursion Protection is a flag that first came from 12th gen onwards and Intel is now recommending this, but once you enable this you lose massive performance on 13th and 14th gen processors, Intel's HW engineers knew the silicon was pushed to max and they gave this failsafe.
Now if you watch Buildzoid videos, most of the Intel Baseline are pushing extreme LLC thus causing higher idle VCore 1.7 (Datasheet since 10th gen shows this very high VCore but IRL 13th and 14th baseline pushes to that high voltage) pumped into the processor to maintain it's stability of 5.7-6Ghz clock rate, if Intel removes those higher speeds there will be a lawsuit so they only have this option to run away and let mobo OEMs take heat, matter of fact is 13th and 14th gen processors are flawed by a significant margin. Stock Cinebench fails, higher voltage and CPU degradation, poor SP rating bins and very bad KS release which has worst variances of binning and high heat.
Add bendgate, poor XMP OC stability, P/E core latency IRL (watch techyescity videos that show lower latency than P core only 10th gen), CPU boost clock stability is proportional to higher voltage and poor binning. Unstable CPUs. BSOD lockups, CPU degradation. 1.7v on 10nm SuperFin and same 1.7v on 14nm++ (10th gen will degrade at this higher voltage)
Intel dropped the ball big time. All those higher scores are now dead if you enable the Intel baseline across mobos.Reply
It depends. Sure the benchmarks might drop, but are they going to drop by 2% (margin of error), or by 5% (notable amount), or by a 10% (substantial degree).
I remember AMD tried to pull a similar thing back with the RX-480 cards, because all the reviewers were using high-end motherboards, so it scored favourably for launch day and early benchmarks. When regular people got them, it had a lot of faults/crashes, because they were using regular motherboards and systems, so the cards couldn't cope. The ironic part is that AMD did eventually concede the loss to Nvidia and it's GTX-1060 only for years later their software and drivers caught up, and time proved the AMD card was actually superior. Same thing with the GTX-680 vs HD7570, GTX-780 vs R9-290, GTX-1070 versus Vega-56, or the RTX-2070-Super versus the RX-5700xt, or the basically most of the 2020 cards RTX 3060, 3060Ti, 3070, 3080Ti versus RX 6700XT, 6800, 6800XT, 6900.
These Corporations try to balance price, marketing, benchmarks, and support. Which is why we get these weird happenings. Reply
Are these recommended settings essentially stock performance, or maximum safe longterm OC levels? At one level the former is what Intel would ideally like their board makers to use, but unless they've shaved away all the safety margin in the last few years the latter is what many enthusiasts want for their daily driver systems. Ideally Intel would provide guidance on how far we can push their systems in relative safety without having to worry about them dying a few months down the line.
The extreme OC record chasers have different desires, but they're also doing stuff they know will frequently kill their stuff anyway and will fiddle with everything regardless of any warnings they need to click past.Reply
BTW, that PL1/PL2/Tau graph is simplistic to the point of being misleading. Tau isn't an absolute number of seconds, but rather a characteristic width of the EWMA (Exponentially-Weighted Moving Average) kernel that's used to compute average power. When the filtered power level exceeds PL1, then the maximum input power drops from PL2 to PL1.
According to this algorithm, if your PL1 is 125 W and your CPU is only using like 150 W, it'll boost for much longer than if it's riding the PL2 limit of 253 W.Reply
Intel 10nm is out of gas and it’s time for a new node and new process ASAP. Raptor lake pushed Intel 10nm to the absolute limit that is the long and short of it.
Hopefully the nodes in the intel EUV era are much more power efficient and the resulting chips (whether from Intel or TSMC) run much more power efficiently, thus leaving more thermal headroom for overlocking. We shall see how arrow lake and the future EUV lakes behave. Reply
The entire situation is comedy gold. The dwindling number of PC enthusiasts are the ones adversely impacted since they're the only ones still willing to effectively waste their future financial well-being on DIY desktop PCs and they're also the ones most likely to unwisely force their computers to do more in the mindless and vacant-headed chase for a couple percent more performance. Meanwhile there are billions of us (literally) that will never, ever be impacted because we just buy a laptop and accomplish useful tasks with it until we need to buy another laptop in 5 to 7 years. Have fun with your exploding, power hog, space heaters nerds!Reply
The growth of "enthusiast' class hardware outstripped other sectors for years now. Dont worry, we know you're jealous as you watch people actually enjoy their PCs, but surely your Mac does everything you need it to do, right :)Reply
Macs? Imagine using a corner case OS with a huge markup to accomplish the same tasks! That would be another variation of the same stupidity - different loyalties but the same moronic brain fault - as PC enthusiast gamer nerds. They need their (near) adult toys as a pacifier, they just happen to have a half-eaten fruit logo on the case.
No thanks bruh. I'm currently typing this on an ex-business laptop purchased used from ebay for $40 USD due to substantial external, yet superficial, case damage. i5-7200U works great but I do tend to linger over my cheap phone and bluetooth keyboard in the summer out of consideration for cooling costs.Reply
"The growth of "enthusiast' class hardware outstripped other sectors for years now."
"Sales of client PCs declined 13.9% year-over-year and totaled 259.5 million units in 2023, according to IDC. Gaming PC shipments experienced a similar downturn as the overall PC market, with a 13.2% year-over-year decrease to 44 million units, analysts from the same firm suggest, which means that gaming machines commanded 17% of the market."
well, not according to the Sister Publication!!! For those too young, or uninformed, the whole PC revolution was not driven by niche applications like Games or CAD/CAM or RDBMS, but by the First Killer App for office workers: Lotus 1-2-3. There's been nothing of that scale since; one might argue that Excel/Office is, but that's just a pretty-pretty version of 1-2-3.
It's easy to "outstrip" if your starting from a very low base. That's what the PC gaming sector is.
The point being, of course, that hitching your wagon to the gaming community will not fill the coffers.Reply
I dunno about all that, you are way over dramatizing here. What we have is Intel, struggling for the past decade with process nodes, and therefore pushing things beyond what they should have to remain competitive.
Even before this situation, anyone buying a Core i9 13 or 14th gen should know they are already being pushed too close to the edge. Buying a Core i9 13 or 14th gen over an AMD Zen 4 seems stupid to me, but... People are. Reply
" Meanwhile there are billions of us (literally) that will never, ever be impacted because we just buy a laptop and accomplish useful tasks with it until we need to buy another laptop in 5 to 7 years " speak for your self, no one i know has a laptop, they are all on desktops. and your idea of " useful tasks " is completely different of someone else's but considering your previous posts were you mention what you use a computer for, and your attitude to any one that doesn't fit your narrative, this type of response, isn't surprising. Reply
Yeah, I have had this same goofy conversation with him before. Kind of makes me wonder who is buying all of those flagship processors. They sure as hell produce alot of Core and Ryzen desktop CPUs for noone. LOL Reply
i just find it hilarious how he thinks HIS usage should dictate how the rest of us should use and buy a comp. while a notebook from 14 years ago suits his needs, i doubt it would suit the needs of most now. his comment " willing to effectively waste their future financial well-being on DIY desktop PC " further shows that point. pure ignorance, and arrogance. while he thinks its a waste, others do not, and are willing to save and buy the comp that fits their needs and usage.Reply
haha, right... "I only need an 8 year old laptop, therefore anything faster is a waste for everyone". Meanwhile desktop chips sales and even newer laptop chip sales carry on without you.
BTW, in that sector,. AMD's mobile Ryzen 7000 and 8000 chips are amazing. Reply
To be fair, the applications that a lot of people use, things like Office and the browser, don't need very modern CPUs. How much Skylake-class hardware is still running in governments and places like that? After Sandy Bridge, CPUs have been reasonably fast and lasting a long time.Reply
GeoffreyA, yes, but it looks like the discussion here with Peach is ones personal computer, that they use, not ones for business and governments, as those do not need a comp made in the last 5 year or so, depending on what they do for work, anyway.Reply
You'll maybe figure out out a few years too late for the change in perspective to make a meaningful difference. People are terrible at understanding the differences between wants and needs and tend to focus on near-term gratification to the detriment to themselves in the future and to this fun little experiment we know as civilization as a whole. But do keep on keeping on and for a while, you'll see how the choices of individuals with such shortsightedness direct the course we collectively take.Reply
"you'll see how the choices of individuals with such shortsightedness direct the course we collectively take"
You sound pretty detached from reality there bud. We are talking about computers here, they are just tools. As far as wants and needs, yours are not mine. I dont "Need" a new laptop but I have one becasue I wanted it. Sure for most of my work an 8 year old i5-7200U would do the job. I do still have an old laptop in that age range in case someone needs it, in fact I think its a i5-6200U. True it will handle general use, web browsing , email office apps etc... But I dont want to keep using it because my time is valuable to me, and I want to get things done faster. I also have a DT for gaming. I know, what a niche market that is. :P Reply
" You'll maybe figure out out a few years too late for the change in perspective to make a meaningful difference " maybe you will as well, then, but i doubt it considering you look down on any one that doesn't conform to YOUR view and narrative.
" People are terrible at understanding the differences between wants and needs and tend to focus on near-term gratification to the detriment to themselves in the future " yes, because every ones wants and needs are vastly different, ever consider that ? sure seems like you dont, and wont. bottom line, while you may not need the latest computer, and as i said, are happy with a notebook thats 14years old, cause dont do any thing that needs anything better, others do. if they can afford it, or save up so they can buy it, how does that cause them any sort of detriment ?
" But do keep on keeping on and for a while, you'll see how the choices of individuals with such shortsightedness direct the course we collectively take." practice what you preach, it sure looks like you are the short sighted one here, as again, you tend to insult any one that doesnt fit your narrative or views when it comes to which computer hardware they buy, which you have shown in quite a few of your posts. Reply
nice try. but no, i was just stating you are the one that look down on any one that doesnt share your view or narrative when it comes to what computer a person has, which you have shown in other posts. the parts of your posts that i quoted above, do just that.
the computer you own and use fits your needs and usage, which is fine. but don't look down on others that chose to get the comps that suits their needsReply
That sure does seem defensive from here. Wasn't even initially talking about you, but just poking a little fun at the people that would waste their time and money on a K-series chip. No one knew you existed until you rushed to their defence to white knight the maybe 2 people on the planet outside of LTT's reviewers that this impacts.Reply
" Wasn't even initially talking about you, but just poking a little fun at the people that would waste their time and money " whille you think its a waste of money, those that can use them, do not, again talking down to people.
what ever peach and cream, keep thinking your better then others, just because you dont have a need for a better computer then you have, while others do.Reply
Most high end chips are not being bought for bragging rights by tech heads, but by people who use software that requires maximum performance - 3d artists, video editors, etc, etc. Maybe it's their hobby, maybe it's for their work, but either way, they need the most powerful CPU they can reasonably afford. That you can't imagine anyone needing ultra fast CPUs speaks to your limited knowledge rather than to any real lack of a need for such chips. Reply
Not really. This does make them look bad, but if they release the next gen stuff and it performs well, and is priced competitively with AMD's linup at hte time people will buy it.
Given Intels track record for the past decade, I give that a very VERY slim chance. We will probably see them kick the can down the road, delaying 6 months, then another 6 months, then another LOL. But it could happen, and if it does, people will buy it. Reply
Or they will allow motherboard makers to do unlimited power again on 15th Gen and wait a year for there to be power related crashes again to acknowledge the issue. However, the benchmarks are already out there and people only look at that and they will laugh to the bank with the sales. Then they will say that RMA's on the CPU isn't allowed because they were run out of spec thus throwing the vendors under the bus.Reply
Shareholders generally care about dividends and stock valuation. If they worry about hiccups involving a halo processor model or two in a micro-segment of the much broader business, then they're probably not the investors that Intel's C-suite would care about since they hold almost no voting power.Reply
powerarmour, I share your sentiment about justice and that this crook company should be brought to book, but don't we all know by now that they get away with almost anything? There is pseudo-religious feeling towards Intel, and these deplorable CPUs won't dent that much.Reply
"we have CPUs shipping with a maximum turbo clock speed of 6.0 GHz and a peak power consumption of over 400 Watts, figures that were only a pipe dream a decade ago."
I'd be more sympathetic towards motherboard manufacturers if they were actually doing any sort of tuning for their default settings. Instead, it's a case of "turn off stability controls, turn all settings up to their maximum (e.g. PL1 at 4096W) and YOLO on CPU survival". The Ryzen 7000 series overvolting and frying themselves should have been a wake-up call across the industry, but it appears Make Number Bigger won out. Reply
Though motherboard manufacturers are immediately to blame, one still feels that Intel is the true culprit, and now that this has happened, they're trying to play the innocent good guy, as if butter wouldn't melt in their mouth.Reply
Intel are mixing RECOMMENDATION and SPECIFICATION disingenuously to their own advantage. The Intel admits even high-end motherboards are "within specification" and that is all that should matter for stability. Your article mixes these two as well.
Intel could choose to make their "so called recommendations" the "actual specifications" - but don't. Why not? It would nerf the performance.
I think your article is not holding Intel to account. And the line "spec-defying state of high-end motherboards " is factually wrong. I would opine a correction is warranted.
Lastly we do not know if silicon degradation has happened, few cases reported are not even the performance impacting "baseline settings" are achieving stability (not verified yet AFAIK).
Anandtech should be "pro-consumer" and not "pro Intel" in both tone and facts.Reply
Ah, i forgot AT has become the gathering point of every AMD fan out there. Intel’s right on this one. The MB vendors are, sometimes, retarded with their OOB “default” settings. Gigabyte has made me mad with their Z170 Aorus Master, which fed 1.5(!!!!!!!!!) Volts to the 7700k. And it happened again with the Z790 and the 13900k. Since i’m used to their shenanigans, i simply spared ~2 hours for a few tests under some voltage offset and i’m ok, now. This would have never happened on a lower-end MB.Reply
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
63 Comments
Back to Article
atragorn - Monday, April 29, 2024 - link
Intel has tacitly approved up until now. Since there is now a hue and a cry they are changing their tune. But lets be real they knew and approved or they would have done something about it a long time ago. Its only the negative press (And possibly Potential RMA concerns ) that cause them to change their tune. Replyshing3232 - Tuesday, April 30, 2024 - link
It worked until weather get hot ReplyGigaplex - Tuesday, April 30, 2024 - link
The weather is always hot somewhere ReplySamus - Tuesday, May 7, 2024 - link
Here in Chicago my solution was to run my 13900k (which was never stable enough to launch Battlefield 2042 and would occasionally greet me with a BSOD otherwise using stock Asus XMPI profile) was a mild vcore offset of +0.050. This obviously increased temperatures in the Winter and wasn't a problem until this last month when it became difficult to keep my room under 80F with my HVAC system while gaming. The RTX 4080 doesn't help, but I measured the CPU produces an extra 100w of heat when using my offset with XMPI (Asus profile) verses stock offset and XMPII (Intel profile) and negligible performance difference.I don't plan to bother running XMPI anymore, period. Whatever performance advantage the Asus profile offered was indistinguishable from Intel's parameters and not worth the hassle. Reply
schujj07 - Tuesday, April 30, 2024 - link
I have thought the same thing. Intel knew and supported unlimited power to CPUs just so they could win benchmarks. However, that has finally caught up to them as 13th and 14th Gen P-Cores are all versions of 12th Gen Golden Cove just pushed to the absolute limit. This what they did with Skylake as well and also why Intel cannot go with 16 P-Cores as power consumption would be astronomical at the clock speeds they need to be competitive. Replydo_not_arrest - Friday, May 3, 2024 - link
You don't know that. You don't actually know anything, you just like to make random statements without bothering to care if they are true or not.Intel has always had more rigorous product validation than any other chip maker. That's a fact. The fact that there are bugs is just the nature of complex machines. Reply
BushLin - Saturday, May 4, 2024 - link
Weird. schujj07 makes a pretty uncontroversial statement. Countered by an unrelated point while behaving unnecessarily rude. Replypowerarmour - Wednesday, May 1, 2024 - link
This 'Intel Baseline Profile' is nothing more than increasing voltages to panic fix stability, shame on them.Oh how the mighty have fallen... Reply
GeoffreyA - Wednesday, May 1, 2024 - link
It's nothing new with them. The Pentium III that broke the GHz barrier, after the Athlon, was unstable and had to be recalled. And early Northwood steppings would degrade when overclocked and at length fail. Replyboozed - Monday, April 29, 2024 - link
The headline is so good that it could easily be satire ReplyOxford Guy - Wednesday, May 1, 2024 - link
I thought it was. Replyeloyard - Tuesday, April 30, 2024 - link
"All benchmarks are done, public perception of performance is established, time to downgrade". ReplyMarcusMo - Tuesday, April 30, 2024 - link
Are you going to re-review the CPUs at these new settings? Feels like intel is trying to have the cake and eat it too here… Replymode_13h - Tuesday, April 30, 2024 - link
Yes, please re-benchmark. Then, we can see just how much of the i9-14900K's rated performance was due to ill-gotten gains. Replypowerarmour - Tuesday, April 30, 2024 - link
Absolutely, let the true performance be shown now. ReplyRyan Smith - Tuesday, April 30, 2024 - link
We haven't decided quite how we'll go about things yet, since that's going to hinge in part on what Intel's official determination in May ends up being - and whether mobo vendors actually change their out-of-the-box defaults or not. But yes, at some point we'll have to pull in some additional benchmark data with the baseline settings. Replykkilobyte - Tuesday, April 30, 2024 - link
You really should re-benchmark. It would be consistent with a statement you regularly made about testing CPUs using JEDEC profiles instead of the XMP ones, because XMP was considered as some sort of out-of-spec overclocking. So shouldn't a similar policy also be applied to other system components, like the CPU? ReplyRyan Smith - Tuesday, April 30, 2024 - link
As things stand, per Intel's warranty terms, the aggressive settings by motherboards are not overclocking the CPU. XMP, on the other hand, explicitly does violate the warranty.Overall, our preference is to test out-of-the-box settings, with the caveat that we won't void warranties in stock testing. So aggressive power settings, which are covered, pass that criteria - though I certainly won't mind if Intel reins in the default settings. Conversely, mobos don't enable XMP by default to begin with, so enabling it would also violate the OOB principle (and even then, I have to admit we'd probably still enable it if Intel's warranty covered it).
Obviously, all of our current settings go out the window if Intel changes the OOB experience. Reply
BushLin - Saturday, May 4, 2024 - link
Just in case it needs pointing out for the millionth time, this site's policy of not enabling XMP in benchmarks, makes the benchmarks useless to its audience. ReplyHresna - Sunday, May 5, 2024 - link
“Useless” is maybe a bit strong - mainly because the benchmarks on the site at least all have an internal consistency amongst themselves (which I’m sure factors heavily in their decision to uphold it), but I agree that it seems a bit antiquated and limiting today.They’ve never claimed papal infallibility, so they could change their mind at some point. In the DDR4 era, it seems 3200Mhz CL16 or 18 would have been the logical choice. Maybe a bit hard to say right now for DDR5, which again probably factors into things.
But given they seem to run this site on a shoestring, I’m kindof just glad they are still in business and there’s other things I’d rather see… like less ad-laden slow-rendering pages and clickbaity-style multi-page delivery. Heck, even in trying to write this comment I had to start over because the safari tab crashed and reloaded the page. Reply
haplo602 - Tuesday, April 30, 2024 - link
What would interest me more is a guide to how we can determine the proper default values and how/where to set them ? F.e. ark.intel does not specify much information on the CPU page. So HOW can I validate that the MOBO maker actually implemented the baseline profile correctly ? ReplySilver5urfer - Tuesday, April 30, 2024 - link
Intel is just throwing their unstable 13th and 14th gen silicon under the rug of OEM BIOS options.Historically speaking as AT already noted Intel PL1 and PL2 and the TAU were there ever since Core series Architecture went mainstay, like since Sandybridge onwards from what I can recall. And Intel Power Limit was never stock and none of them failed like these.
Now looking at Buildzoid videos and other forum talk (OCN, Notebooktalk), Intel pushed these processors to extreme for their marketing team so that they can upsell these despite higher heat density and power consumption which is far higher than Intel 11th gen Rocketlake 14nm backport. 13th gen is impossible to cool on an AIO, whereas 11th can barely manage, 10th can do it. And 13th / 14th gen are extremely hot on even an AIO, the reasons are - Intel 6GHz marketing, their new Ring OC pushed beyond 5GHz (10th gen can OC past 5GHz but they had infamous WHEA issues, only way to fix was either get lucky by silicon lottery top bin of SP rating or you degrade the cache) and XMP.
Now the 13th / 14th gen processors binning is atrocious, for this reason many processors cannot finish Cinebench R15-R23. Due to the silicon instability out of the box, generally speaking most of the silicon is given higher voltage to cope up with the mean distribution of avg voltage across the CPU arch, and still that is not enough. Until now Intel Core processors never had such instability issues OOTB. Stock or MCE, they never exhibited this extreme instability, and matter of fact Intel CEP, Current Excursion Protection is a flag that first came from 12th gen onwards and Intel is now recommending this, but once you enable this you lose massive performance on 13th and 14th gen processors, Intel's HW engineers knew the silicon was pushed to max and they gave this failsafe.
Now if you watch Buildzoid videos, most of the Intel Baseline are pushing extreme LLC thus causing higher idle VCore 1.7 (Datasheet since 10th gen shows this very high VCore but IRL 13th and 14th baseline pushes to that high voltage) pumped into the processor to maintain it's stability of 5.7-6Ghz clock rate, if Intel removes those higher speeds there will be a lawsuit so they only have this option to run away and let mobo OEMs take heat, matter of fact is 13th and 14th gen processors are flawed by a significant margin. Stock Cinebench fails, higher voltage and CPU degradation, poor SP rating bins and very bad KS release which has worst variances of binning and high heat.
Add bendgate, poor XMP OC stability, P/E core latency IRL (watch techyescity videos that show lower latency than P core only 10th gen), CPU boost clock stability is proportional to higher voltage and poor binning. Unstable CPUs. BSOD lockups, CPU degradation. 1.7v on 10nm SuperFin and same 1.7v on 14nm++ (10th gen will degrade at this higher voltage)
Intel dropped the ball big time. All those higher scores are now dead if you enable the Intel baseline across mobos. Reply
GeoffreyA - Tuesday, April 30, 2024 - link
If I'm not mistaken, the last there was a problem similar to this was in the early Northwood steppings. ReplyKangal - Sunday, May 5, 2024 - link
It depends.Sure the benchmarks might drop, but are they going to drop by 2% (margin of error), or by 5% (notable amount), or by a 10% (substantial degree).
I remember AMD tried to pull a similar thing back with the RX-480 cards, because all the reviewers were using high-end motherboards, so it scored favourably for launch day and early benchmarks. When regular people got them, it had a lot of faults/crashes, because they were using regular motherboards and systems, so the cards couldn't cope. The ironic part is that AMD did eventually concede the loss to Nvidia and it's GTX-1060 only for years later their software and drivers caught up, and time proved the AMD card was actually superior. Same thing with the GTX-680 vs HD7570, GTX-780 vs R9-290, GTX-1070 versus Vega-56, or the RTX-2070-Super versus the RX-5700xt, or the basically most of the 2020 cards RTX 3060, 3060Ti, 3070, 3080Ti versus RX 6700XT, 6800, 6800XT, 6900.
These Corporations try to balance price, marketing, benchmarks, and support. Which is why we get these weird happenings.
Reply
DanNeely - Tuesday, April 30, 2024 - link
Are these recommended settings essentially stock performance, or maximum safe longterm OC levels? At one level the former is what Intel would ideally like their board makers to use, but unless they've shaved away all the safety margin in the last few years the latter is what many enthusiasts want for their daily driver systems. Ideally Intel would provide guidance on how far we can push their systems in relative safety without having to worry about them dying a few months down the line.The extreme OC record chasers have different desires, but they're also doing stuff they know will frequently kill their stuff anyway and will fiddle with everything regardless of any warnings they need to click past. Reply
mode_13h - Tuesday, April 30, 2024 - link
BTW, that PL1/PL2/Tau graph is simplistic to the point of being misleading. Tau isn't an absolute number of seconds, but rather a characteristic width of the EWMA (Exponentially-Weighted Moving Average) kernel that's used to compute average power. When the filtered power level exceeds PL1, then the maximum input power drops from PL2 to PL1.According to this algorithm, if your PL1 is 125 W and your CPU is only using like 150 W, it'll boost for much longer than if it's riding the PL2 limit of 253 W. Reply
Exotica - Tuesday, April 30, 2024 - link
Intel 10nm is out of gas and it’s time for a new node and new process ASAP. Raptor lake pushed Intel 10nm to the absolute limit that is the long and short of it.Hopefully the nodes in the intel EUV era are much more power efficient and the resulting chips (whether from Intel or TSMC) run much more power efficiently, thus leaving more thermal headroom for overlocking. We shall see how arrow lake and the future EUV lakes behave. Reply
PeachNCream - Tuesday, April 30, 2024 - link
The entire situation is comedy gold. The dwindling number of PC enthusiasts are the ones adversely impacted since they're the only ones still willing to effectively waste their future financial well-being on DIY desktop PCs and they're also the ones most likely to unwisely force their computers to do more in the mindless and vacant-headed chase for a couple percent more performance. Meanwhile there are billions of us (literally) that will never, ever be impacted because we just buy a laptop and accomplish useful tasks with it until we need to buy another laptop in 5 to 7 years. Have fun with your exploding, power hog, space heaters nerds! ReplyTheinsanegamerN - Tuesday, April 30, 2024 - link
The growth of "enthusiast' class hardware outstripped other sectors for years now. Dont worry, we know you're jealous as you watch people actually enjoy their PCs, but surely your Mac does everything you need it to do, right :) ReplyPeachNCream - Tuesday, April 30, 2024 - link
Macs? Imagine using a corner case OS with a huge markup to accomplish the same tasks! That would be another variation of the same stupidity - different loyalties but the same moronic brain fault - as PC enthusiast gamer nerds. They need their (near) adult toys as a pacifier, they just happen to have a half-eaten fruit logo on the case.No thanks bruh. I'm currently typing this on an ex-business laptop purchased used from ebay for $40 USD due to substantial external, yet superficial, case damage. i5-7200U works great but I do tend to linger over my cheap phone and bluetooth keyboard in the summer out of consideration for cooling costs. Reply
FunBunny2 - Thursday, May 2, 2024 - link
"The growth of "enthusiast' class hardware outstripped other sectors for years now.""Sales of client PCs declined 13.9% year-over-year and totaled 259.5 million units in 2023, according to IDC. Gaming PC shipments experienced a similar downturn as the overall PC market, with a 13.2% year-over-year decrease to 44 million units, analysts from the same firm suggest, which means that gaming machines commanded 17% of the market."
-- https://www.tomshardware.com/tech-industry/gaming-...
well, not according to the Sister Publication!!! For those too young, or uninformed, the whole PC revolution was not driven by niche applications like Games or CAD/CAM or RDBMS, but by the First Killer App for office workers: Lotus 1-2-3. There's been nothing of that scale since; one might argue that Excel/Office is, but that's just a pretty-pretty version of 1-2-3.
It's easy to "outstrip" if your starting from a very low base. That's what the PC gaming sector is.
The point being, of course, that hitching your wagon to the gaming community will not fill the coffers. Reply
goatfajitas - Tuesday, April 30, 2024 - link
I dunno about all that, you are way over dramatizing here. What we have is Intel, struggling for the past decade with process nodes, and therefore pushing things beyond what they should have to remain competitive.Even before this situation, anyone buying a Core i9 13 or 14th gen should know they are already being pushed too close to the edge. Buying a Core i9 13 or 14th gen over an AMD Zen 4 seems stupid to me, but... People are. Reply
Qasar - Tuesday, April 30, 2024 - link
" Meanwhile there are billions of us (literally) that will never, ever be impacted because we just buy a laptop and accomplish useful tasks with it until we need to buy another laptop in 5 to 7 years "speak for your self, no one i know has a laptop, they are all on desktops. and your idea of " useful tasks " is completely different of someone else's but considering your previous posts were you mention what you use a computer for, and your attitude to any one that doesn't fit your narrative, this type of response, isn't surprising. Reply
goatfajitas - Wednesday, May 1, 2024 - link
Yeah, I have had this same goofy conversation with him before. Kind of makes me wonder who is buying all of those flagship processors. They sure as hell produce alot of Core and Ryzen desktop CPUs for noone. LOL ReplyQasar - Wednesday, May 1, 2024 - link
i just find it hilarious how he thinks HIS usage should dictate how the rest of us should use and buy a comp. while a notebook from 14 years ago suits his needs, i doubt it would suit the needs of most now. his comment " willing to effectively waste their future financial well-being on DIY desktop PC " further shows that point. pure ignorance, and arrogance. while he thinks its a waste, others do not, and are willing to save and buy the comp that fits their needs and usage. Replygoatfajitas - Wednesday, May 1, 2024 - link
haha, right... "I only need an 8 year old laptop, therefore anything faster is a waste for everyone". Meanwhile desktop chips sales and even newer laptop chip sales carry on without you.BTW, in that sector,. AMD's mobile Ryzen 7000 and 8000 chips are amazing. Reply
GeoffreyA - Wednesday, May 1, 2024 - link
To be fair, the applications that a lot of people use, things like Office and the browser, don't need very modern CPUs. How much Skylake-class hardware is still running in governments and places like that? After Sandy Bridge, CPUs have been reasonably fast and lasting a long time. ReplyQasar - Wednesday, May 1, 2024 - link
GeoffreyA, yes, but it looks like the discussion here with Peach is ones personal computer, that they use, not ones for business and governments, as those do not need a comp made in the last 5 year or so, depending on what they do for work, anyway. ReplyPeachNCream - Wednesday, May 1, 2024 - link
You'll maybe figure out out a few years too late for the change in perspective to make a meaningful difference. People are terrible at understanding the differences between wants and needs and tend to focus on near-term gratification to the detriment to themselves in the future and to this fun little experiment we know as civilization as a whole. But do keep on keeping on and for a while, you'll see how the choices of individuals with such shortsightedness direct the course we collectively take. Replygoatfajitas - Wednesday, May 1, 2024 - link
"you'll see how the choices of individuals with such shortsightedness direct the course we collectively take"You sound pretty detached from reality there bud. We are talking about computers here, they are just tools. As far as wants and needs, yours are not mine. I dont "Need" a new laptop but I have one becasue I wanted it. Sure for most of my work an 8 year old i5-7200U would do the job. I do still have an old laptop in that age range in case someone needs it, in fact I think its a i5-6200U. True it will handle general use, web browsing , email office apps etc... But I dont want to keep using it because my time is valuable to me, and I want to get things done faster. I also have a DT for gaming. I know, what a niche market that is. :P Reply
Qasar - Wednesday, May 1, 2024 - link
" You'll maybe figure out out a few years too late for the change in perspective to make a meaningful difference "maybe you will as well, then, but i doubt it considering you look down on any one that doesn't conform to YOUR view and narrative.
" People are terrible at understanding the differences between wants and needs and tend to focus on near-term gratification to the detriment to themselves in the future "
yes, because every ones wants and needs are vastly different, ever consider that ? sure seems like you dont, and wont. bottom line, while you may not need the latest computer, and as i said, are happy with a notebook thats 14years old, cause dont do any thing that needs anything better, others do. if they can afford it, or save up so they can buy it, how does that cause them any sort of detriment ?
" But do keep on keeping on and for a while, you'll see how the choices of individuals with such shortsightedness direct the course we collectively take."
practice what you preach, it sure looks like you are the short sighted one here, as again, you tend to insult any one that doesnt fit your narrative or views when it comes to which computer hardware they buy, which you have shown in quite a few of your posts. Reply
PeachNCream - Thursday, May 2, 2024 - link
"maybe you will as well, then, but i doubt it considering you look down on any one that doesn't conform to YOUR view and narrative."The entirety of your reply was exactly that. Reply
Qasar - Saturday, May 4, 2024 - link
nice try. but no, i was just stating you are the one that look down on any one that doesnt share your view or narrative when it comes to what computer a person has, which you have shown in other posts. the parts of your posts that i quoted above, do just that.the computer you own and use fits your needs and usage, which is fine. but don't look down on others that chose to get the comps that suits their needs Reply
PeachNCream - Saturday, May 4, 2024 - link
That sure does seem defensive from here. Wasn't even initially talking about you, but just poking a little fun at the people that would waste their time and money on a K-series chip. No one knew you existed until you rushed to their defence to white knight the maybe 2 people on the planet outside of LTT's reviewers that this impacts. ReplyQasar - Saturday, May 4, 2024 - link
" Wasn't even initially talking about you, but just poking a little fun at the people that would waste their time and money " whille you think its a waste of money, those that can use them, do not, again talking down to people.what ever peach and cream, keep thinking your better then others, just because you dont have a need for a better computer then you have, while others do. Reply
Glaurung - Wednesday, May 1, 2024 - link
Most high end chips are not being bought for bragging rights by tech heads, but by people who use software that requires maximum performance - 3d artists, video editors, etc, etc. Maybe it's their hobby, maybe it's for their work, but either way, they need the most powerful CPU they can reasonably afford. That you can't imagine anyone needing ultra fast CPUs speaks to your limited knowledge rather than to any real lack of a need for such chips. Replypowerarmour - Tuesday, April 30, 2024 - link
Intel are done in the enthusiast market unless they take full ownership of this mess and admit they screwed up. Replygoatfajitas - Tuesday, April 30, 2024 - link
Not really. This does make them look bad, but if they release the next gen stuff and it performs well, and is priced competitively with AMD's linup at hte time people will buy it.Given Intels track record for the past decade, I give that a very VERY slim chance. We will probably see them kick the can down the road, delaying 6 months, then another 6 months, then another LOL. But it could happen, and if it does, people will buy it. Reply
schujj07 - Tuesday, April 30, 2024 - link
Or they will allow motherboard makers to do unlimited power again on 15th Gen and wait a year for there to be power related crashes again to acknowledge the issue. However, the benchmarks are already out there and people only look at that and they will laugh to the bank with the sales. Then they will say that RMA's on the CPU isn't allowed because they were run out of spec thus throwing the vendors under the bus. ReplyGeoffreyA - Tuesday, April 30, 2024 - link
People will still support Intel no matter what they do. Replypowerarmour - Tuesday, April 30, 2024 - link
Shareholders won't ReplyPeachNCream - Tuesday, April 30, 2024 - link
Shareholders generally care about dividends and stock valuation. If they worry about hiccups involving a halo processor model or two in a micro-segment of the much broader business, then they're probably not the investors that Intel's C-suite would care about since they hold almost no voting power. ReplyGeoffreyA - Wednesday, May 1, 2024 - link
powerarmour, I share your sentiment about justice and that this crook company should be brought to book, but don't we all know by now that they get away with almost anything? There is pseudo-religious feeling towards Intel, and these deplorable CPUs won't dent that much. Replyjamesindevon - Tuesday, April 30, 2024 - link
"we have CPUs shipping with a maximum turbo clock speed of 6.0 GHz and a peak power consumption of over 400 Watts, figures that were only a pipe dream a decade ago."More like a pipe nightmare. Reply
edzieba - Tuesday, April 30, 2024 - link
I'd be more sympathetic towards motherboard manufacturers if they were actually doing any sort of tuning for their default settings.Instead, it's a case of "turn off stability controls, turn all settings up to their maximum (e.g. PL1 at 4096W) and YOLO on CPU survival". The Ryzen 7000 series overvolting and frying themselves should have been a wake-up call across the industry, but it appears Make Number Bigger won out. Reply
Belmet - Tuesday, April 30, 2024 - link
https://www.anandtech.com/show/14582/talking-tdp-t... ReplyGeoffreyA - Tuesday, April 30, 2024 - link
Though motherboard manufacturers are immediately to blame, one still feels that Intel is the true culprit, and now that this has happened, they're trying to play the innocent good guy, as if butter wouldn't melt in their mouth. ReplyOxford Guy - Saturday, May 11, 2024 - link
'Though motherboard manufacturers are immediately to blame'I don't agree. It takes two to tango in this case. Reply
GeoffreyA - Wednesday, May 15, 2024 - link
Agreed. It's both. ReplyTekCheck - Tuesday, April 30, 2024 - link
Blaming vendors and silently supporting high bencharks is ugly, indeed... Replyjayfang - Monday, May 6, 2024 - link
Intel are mixing RECOMMENDATION and SPECIFICATION disingenuously to their own advantage. The Intel admits even high-end motherboards are "within specification" and that is all that should matter for stability. Your article mixes these two as well.Intel could choose to make their "so called recommendations" the "actual specifications" - but don't. Why not? It would nerf the performance.
I think your article is not holding Intel to account. And the line "spec-defying state of high-end motherboards " is factually wrong. I would opine a correction is warranted.
Lastly we do not know if silicon degradation has happened, few cases reported are not even the performance impacting "baseline settings" are achieving stability (not verified yet AFAIK).
Anandtech should be "pro-consumer" and not "pro Intel" in both tone and facts. Reply
liquid_c - Thursday, May 9, 2024 - link
Ah, i forgot AT has become the gathering point of every AMD fan out there.Intel’s right on this one. The MB vendors are, sometimes, retarded with their OOB “default” settings. Gigabyte has made me mad with their Z170 Aorus Master, which fed 1.5(!!!!!!!!!) Volts to the 7700k. And it happened again with the Z790 and the 13900k. Since i’m used to their shenanigans, i simply spared ~2 hours for a few tests under some voltage offset and i’m ok, now. This would have never happened on a lower-end MB. Reply
Oxford Guy - Thursday, May 9, 2024 - link
'Ah, i forgot AT has become the gathering point of every AMD fan out there.'Ad hominem is a fallacy. Reply