Gotta love how the "intel baseline" isn't the same across motherboard vendors. Almost like the intel baseline isn't provided by intel but cooked up by each vendor on their own. Also it's worth noting that SVID set to intel fail safe runs more Vcore than the ASUS default settings for any scenario that doesn't hit the power limit. Gigabyte's intel baseline profile does the same. Most gigabyte boards use AC_LL and DC_LL of less than 1.1mOhms on default but with the baseline profile the AC and DC_LL get set to 1.7mOhm which raise the core voltage for any workload that doesn't hit power limits.
It almost feels like these "Intel baselines" were something nobody cared about for years, so much so, that now mb manufacturers are really scratching their heads. "Wait, what, there are Intel baselines? Who says?" "Dude, Intel, of course, who else. They're blaming us for not abiding by them." "Oh-oh. Ok, let's fix it at once then. Give me the numbers." "Yeah, about that... Thing is, we are not sure we have the numbers. Trying to contact Intel to provide us with them, but they are keeping us on hold for several days. The last guy said he'll go to look in their archives and the one before that said he'll consult the engineers and we haven't heard from either of them since". 😁
The part that bugs me is that these kind of thing have been a rule and not an exception for a while and there seems that there haven't been blatant issues up until now. This bugs me because it narrows down the potential issues to be likely something along the line of a bad batch or suppliers of materials or even worse a known design flaw somewhere within the chip. I can also get why they would be cagey about it because of the possibility of some kind of class action involved as well. If you tell people they are buying "top of the line" type parts and then they end up with the same performance as a middle of the range I7 (Don't quote me on that, I haven't checked it up again) due to how power limits magically need to be enforced now that is going to make a lot Intel buyers start frothing at the mouth. Also it means we should probably rename the Intel baseline specs to "should-have-bought-AMD-mode". I know I am being really cynical and sarcastic but this is going to have serious consequences for some people. Lasty , dumb question has been bugging me. Has anyone seen any positive or negative correlation between cooler power and the issues with these chips? I cant figure out if having better or worse temperatures would mean more wear on the CPU in this situation. Is lower temperature but higher voltage equal or worse than higher temperature lower voltage and throttling for these CPUs?
Funny when i put defaults on motherboard msi carbon z790 thats when windows crashed however when i put a balanced OC no crashes since didnt mess w the long / short duration should i be worried? Temps are in check too
@@shiraz1736 Why not? My 14700k is a very strong performer across all uses. And in gaming I like the higher minimum 1%/0.1% FPS I get vs. AMD on the same games.
@@shiraz1736I bough intel even knowing that thers problem long time ago with motherboards overclocking cpu in default bios settings . It's strange to not check all the info on internet before buying. And on asus motherboard is only one option to click in bios to change it. And still for me intel won. I was choosing between 7950x3d and I 13900 k and Intel was cheaper, same motherboard was cheaper and I know it that any ram I will buy will work. And I know that on amd only 8 cores have cash and I didn't want to play with core scheduling for every application that I can use.
Intel: Motherboard manufacturers, please make sure we win in benchmarks Also Intel: Motherboard manufacturers are causing the crashes on your systems by running out of spec
@@johnd.1618 Stop you are not helping AMD or its users with this kind of comments, acting like a fanboy is irrational, neither AMD or Intel are ours friends they want our hard won money and that's it. And if you are wondering, I'm a AMD Ryzen user.
@@johnd.1618just because most people don’t think on as low as a level as you do doesn’t mean we dint understand. We just understand so much more than you could. Ryzen user for years here.
@@johnd.1618Steve specifically mentions they've been bringing up this in-spec/out-of-spec nonsense for years, even prior to 13th and 14th gen. It's only now that power delivery's gotten so out of hand that the chickens have come home to roost and at an accelerated pace. HUB and other tech outlets have been mentioning the writing on the wall for well over two years. What're you talking about?
UserBenchmark: "The Intel Core 14900K is a 24 Core processor, and the fastest CPU in the world. It has a massive 6GHz boost clock, which is much higher than the Ryzen 7950X3D's boost clock of 5.7GHz. Gamers or Creators will have very little reason to look at the 7950X3D or the 7950X due to it costing $100 and performing roughly 10% in single-core and multi-core. However, an army of AMD's idiot baboon fanboys like Gamer's Nexus and Hardware unboxed will still try to convince you that the 7950X3D is actually worth it, and claim that Intel's 14900K experiences 'stability issues', even though that is only typically the result of extreme high overclocks. Even with the baseline profile, the 14900K beats the 7950X3D in single-core, which further proves the worthlessness of AMD's processor, and the fact that Ryzen CPUs are no better than Bulldozer CPUs. An average user or gamer should not look at any of AMD's processors, as doing so will only result in disappointment due to lower performance, more instability and worse software."
@@christophermullins7163 How ? did they like market that we're proud of CPUs exploding and then Intel got into FOMO mode and said we gotta burn our CPUs too ? lol
The best proof that power limits on Intel are a horrible mess is that even 'Intel baseline' profiles are different. It's great that - as usually - you try to get to the bottom of it instead of believing Intel marketing trying to deflect blame.
Overclocking used to be a deliberate choice - one for people willing to risk stability and deal with higher power, for marginal gains. But these days, out-of-factory overclocking seems to be the new default, and you need to deliberately go and limit everything yourself to get back a cool and stable system which performs pretty much the same...
In the past it wasn't rare for manufacturers to play so safe with default base performance that it was feasible to get about 10% performance increase without doing much to efficiency. GPUs such as the HD 7950 allowed pretty much every owner to increase performance by a higher % than you can get nowadays changing a 4060 for a 4060ti
@@Rentta I guess it depends how recent is "recent", and how marginal is "marginal". Something like 2600k was more than ten years ago. What's the most recent mainstream component that was still decently overclockable?
@@RenttaOverclocking was usually in the neighbourhood of 10% gains, which IMO is somewhat marginal. Is 66fps that good if previously you had 60fps? I'd wager that no, not really. You need significantly higher performance if you want to feel a difference.
Remember the old days when processors had so much overclocking headroom or you could even re-enable disabled cores in some cases? Now they run them right at the edge of stability, or in Intel's case not even that
You can still enable and disable cores relatively easy, with windows and CPU vendors tools, in case of AMD with Ryzen master app for example. But this can also be a problem, because people forget stuff, I know one person that disabled cores, then forgot about it, latter made a angry youtube video saying that Microsoft windows was dumb, because "it didn't saw all the cores of his CPU!", when in reality like I said he manually disabled the cores and just forgot...
@@Argoon1981 true, but I'm talking about old 3 or 2 core processors that were sold that way but actually had 4 usable cores in some cases. I think there were even 4 core Phenom IIs that could be unlocked to 6-core
i swear the new amd chips come out see 95c thermal profile...ppl gasped...etc....then quickly realized it could be changed to 75c....and still equate GREAT workload performance and equally great in gaming/shared loads. What a circle back in usage.....meanwhile trying to uv a "fresh and new" intel or amd laptop on the otherhand....those things are rigged to blow and marketed to be "upgraded" on a constant basis to their PURE 'laptop user" demographic its a sad state of affairs. Granted huge leaps in performance on the mobile front in the last 5 years...in realworld uses and theoreticals. The overall thermals/powerprofile aspects still has me largely thinking this gen is worse from a stability standpoint and system health over the course of ownership. And ive built unlocked phenom x2 DFI lanparty rigs in the past...even hybrid crossfire rigs with APU and overclocked chipsets....weird how theres more headscratching involved and general system/degredation in play these days if you want to tune a system or PUSH a budget system to its limits(in the past) considering you cant even really do somthing on par with a "buffed" budget build anymore.
You can be absolutely sure the OEMs like HP, Lenovo and Dell are being told: "You better fckn set the values we 'recommend' on your professional lines! What you do on the consumer crap is none of our business, but if the professional lines aren't stable we will crucify you with a jackhammer!".
@@DragonOfTheMortalKombat Oh they are fully going towards that. Intel had a clever idea of putting in smaller E cores for power efficacy like mobile phones had and while it's a great idea for laptop when you pair 8 E core with 2 P cores. It's basically worthless for desktop when even after windows added support for it the E cores largely don't do much and because of the whole E cores idea the P cores (while powerful) are super inefficient. Just like how bulldozer had a great idea of boosting multithreading at the expense of single threading. Intel is in the middle of needed to double down as a architecture is designed to last a decade or more phase and you can't just turn the ship that quickly. So they're dumping tons of power in it to compensate. I really like the big little idea but for Intel the big cores just lose to Zen and the little cores aren't good enough to replace a multithread so they aren't really needed.
“I used to be in spec, but then they changed what in spec was. Now what I’m with isn’t in spec anymore and what’s in spec seems weird and scary. It’ll happen to you!” -Intel, or motherboard manufacturers. Or Grandpa Simpson.
I don't know if it's silicon, because we know there is heatsink contact problems with the 1700 pin stuff, so I would be more inclined to say it's thermal paste degradation.
@@SidneyCritic truth is we don't know exactly. but if thermals were a problem, wouldn't the cpu just lower the clocks instead of crashing? crashing means the load, clocks and voltages are no (longer) ok. one (of many, granted) probability is silicon degradation.
@@SidneyCritic It's the silicon. Degradation causes the CPU to run a workload with instability, forcing you to undervolt or change its clock speed to continue working stably. It's the same thing that happens if you overvolt a CPU and degrade it, you can no longer run it at the same clock speeds and settings without it being unstable.
@@riven4121I guess you meant overvolt instead of undervolt? After degradation takes place, the CPU will require a higher voltage to mantain the same clock speed as before.
Agree. AMD did a really good job for the 7800X3D disaster. They reacted to the burnt issue very quickly, admitted the mistake and presented the fix before it's getting worse. Now, no 7800X3D will burn and it's still seating in the crown. While Intel just keeps finding excuses 🤐 and being bro with Userbenchmark.
I don't think userbenchmark has anything to do with Intel and also AMD wasn't really at fault for exploding CPUs, mobo manufacturers were with their insanely high memory control voltages.
@@johnd.1618 That's not the point. The point is that even Intel doesn't really approve of userbenchmark. That side is run by someone who most likely has mental issues.
When Mobo vendors did ignoore AMDs set fail safe guidelines and disabled fail safes, AMD shut them down by enforcing it with a new firmware. as intel should have done years ago.
Kept going back and forth with my new build. Ended up switching to AMD for the first time in about 15 years. Decided to get a Ryzen 7800 X3D, with a RTX 4080 super.
Don't forget to check out how to use AMD's Curve Optimiser in the BIOS to moderately undervolt your CPU and get extra performance out of it. I'm running my 7800X3D at -27. Some people can do even better but beware benchmarks that rely only on Cinebench as that's small enough to fit in the L3 cache and doesn't stress the memory subsystem. Even so, the irony that not only are AMD CPUs far more efficient than Intel's to begin with but that you can tweak them to be even more efficient to get more performance versus Intel's approach of throwing voltage at everything is too delicious for words.
@@SOMEONE23145 Thanks, that's exactly what I was thinking as well. I went with the 7800, thinking I could upgrade in a year or two. I got the MSI X670E GAMING PLUS
The 'where can I Blu-Tac a processor to for B-Roll?' game is pushing the same level of your continuing effort to highlight anti-consumer practice and I remain here for it!
@@BigFoot47-48I've had a 7800X3D since September of last year. I've had no issues. I concur with Josh. It's easy to keep cool. I have an Deep Cool AK620 Air Cooler. The upgrade from my 5800X was substantial.
@@BigFoot47-48 Run hardware info 64 and run your games & watch vsoc like a hawk....I undervolted pbo disabled eco mode the works & helldivers 2 would spike to 2.2v and hit 141 c ! granted it was for a split second but the repeatability promted a return. newest bios can help it depends and varies.
9:25 I can confirm that this has to be a bug in the Gigabyte BIOS. I updated to the latest BIOS and saw the same issue (not with an i9, though). Boost behaviour was completely broken and especially the E-cores didn't boost *at all* for any period of time. I saw a performance loss of ~30% with these settings. After reverting back to the previous BIOS version and loading my own baseline profile (where I basically only set the power- and current limits to the "recommended" Intel values), everything went back to normal and for some odd reason performance even increased slightly, while temps and power draw were fine.
@@zodwraith5745 The reason under-volting increases performance is because it is no longer thermal throttling. Thermal throttling isn't binary like the old days; the speed ramp is relative to delta temperature and rate of temperature rise above the threshold. I also have an i7 and under-volted it from stock to 1.2V. It OC no problems to 5 GHz on air, and has run this way for 7 years.
SI here, we sell prebuilds with i9-13900K and I9-14900K. Mid range models with b760 boards that stick to Intel stock settings. And two 14900K failed with us was on stock limits. I have video proof as we were baffled that the degradation happened at stock power. We have had 13900k fail on b760.
@josephlai99 As shared above, i am working for an SI. So bios is our main game. In case you didn't notice, i am talking about a b760 chipset board that doesn't allow overclocking. Yes, brands like Asus enable Asus performance enhancement by default on all Asus boards. But this is disabled in the midrange options. Every system is stress tested and benchmarked while measuring power and temperature as part of our quality control process. So 253 watts, that's all that goes in under any condition. Ofcourse we have higher end models using rog strix b760 and z790. These boards with asus performance enhancement enabled pulls upto 390watts peak. We had failures in stock power.
Really...you should never trust "default" or "stock" settings, ever. I have ALWAYS checked the defaults to ensure they are reasonable. A power limit of 4 kW is NOT reasonable. It would raise a huge red flag for me, and cause me to set the power limits to whatever the spec sheet defined. 253W is a very high figure still, given that until now, most CPU packages have been limited to 90-120W due to thermals. YOU CAN'T CHANGE THE LAWS OF PHYSICS, but you can sacrifice your device for short-term performance. It's like drag racing: you can take a 5.7L V8 and make it produce 4000 HP, but it can only run 1/4 mile before it needs a total rebuild.
I think he had to admit 7800x3d is the better gaming CPU some time ago even with all that min-maxing of Intel CPUs and die cooling, and internal hate for anything with AMD on it (not sure what his problem is).
@@andreiga76 Well we have also never figured out what Userbenchmarks' problem is. Maybe they both don't like a female CEO or the colour red ? who knows lmao
Its pretty shady that Intel threw the motherboard manufacturers under the bus in their statement. They allowed them to do that the entire time (and may have even encouraged it).
They indeed _have_ been encouraging it or at least conveniently overlooked it for more than half of the last decade. _Since at the end of the day, the _*_only_*_ lone beneficiary is Intel itself_ … I mean, didn't Asus engineered their M.C.E in collaboration with Intel itself?! *They deliberately skewed benchmarks ever since and freely allowed reviewers to run **-out of spec-** **_in spec_** and thus, glossed over it as being totally fine.* Meanwhile reviewers (pretty much all of them!) shadily were happily hiding behind the alleged delivery-condition and every question regarding power-draw, heat-dissipation and whatnot, was disregarded as running just the usual 'out of the box' experience, when in fact *_everyone_*_ involved in this mysterious Bermuda triangle of Intel, OEMs and reviewers _*_knew very well, that *exactly nothing was ever actually running at official specifications_* ... Yet no-one dared to care but held open their hands for their share of hush-money (OEMs got their rebates, while reviewers are paid in cost-free samples). All this is done since ages on the back of the clueless customer and his pockets of quicker dying parts and his power-bill, while never ever going to get what he actually paid for, since it's pretty much non-existing. Since if Intel's parts were running at actual specs, Intel-CPUs would rate a very poor second, while being still power-hungry, inefficient and a hot-mess.
The best part of this is Intel was happy with the increased power limits on Publications because it gave them good performance in reviews on release. They've been doing this for years and it finally caught up to them. AMD is doing something similar with PBO and increased power limits but not nearly as severe. All these extra default boosts on motherboards are why people can't get much of an overclock because it's running beyond the default specs out of the box.
It wasn't an issue with those older CPU's because the voltage stayed within the higher end of reason. Now the VCore spikes up +1.5v on these processors. The old rule of thumb for a 24/7 overclock was to keep the VCore below 1.35v. Intel has deluded themselves, vendors, and noobies into thinking 1.45v to 1.55v is safe, when it isn't and never was for long term stability.
So long story short, without those baseline limiters to keep the CPU healthy and adjusting for silicon lottery, AMD's cpus are technically faster than Intels all these years. AMD cpus run cooler and at half the power meaning that they have headroom for overclocking
@@m8x425 That's not necessarily true, it depends on silicon quality, masking quality, and how well the processors are designed internally to regulate voltage and prevent fall-off. Over time, you would generally expect safe v-core voltages to increase, particularly as processors get better at having consistent voltage across all transistors.
@@LupusAries Wow, you are helping so much, when you speak 3 words in slang. I googled your words and also specifically just "Palps". Didn't know, i wanted to learn more about insects.
@@tobytoxd;) I didn't want to spoil it by saying it was about "creamy Sheev" Palpatine/Imperator Palpatine.....there is a rather infamous scene where he goes "UNLIMITED POWAAAH!!!" ;)
@@LupusAries Ah, thanks! That's ensaberlightening to say the least. As a German i have to admit, i did watch those movies in that germanic accent, which didn't end up as famous as the original. SHAME! .)
This is why I like Hardware Unboxed; they always remember. I saw a few outlets and youtubers making this a "new problem", when I remember my Z170 ASUS Maximus VIII Hero (Skylake CPU) shipping with MCE enabled by default already. //edited to correct Intel Chipset (Z170)
@@disco.volantemaybe not, but maybe motherboards pushed CPUs a little too far when the power limit was removed and when the CPU needs more voltage for a given frequency, it just gives up, can't handle that and there you go, instability
@@prosecanlik4296 Given how high the frequencies were, Intel was going to run into this problem sooner or later. Intel 7 node was initally made for the 5.2 Ghz 12900K. That some node has been used for a 6.2Ghz processor. Any good overclocker will tell you that this much increase in clockspeed is simply not sustainable for everyday use. It is way too unstable unless you increase voltage a lot which is harmful to the processor. On top of that, all these CPUs will happily go to 100C before downclocking or throttling.
IDK mate, the one's to blame is surely AMD. AMD's parts being so power efficient made intel pursue the same performance without the efficiently, causing all this stability issue. Without AMD we would still enjoy super stable intel CPUs, which might be a little slower and lack some half the cores.
Since 12 gen they are just cranking up the power to get a perception of better performance. Since 13th gen i thought that it couldn't be infinity and the next architecture should have problemas with it (14th gen uses the same).
Yes I am curious how the next gen if CPU will fare. They are already removing hyperthreading to get more performance and should use lower power by removing some instructions. However will that be emough?
the sheer amount of tinkering/work b-die memory/tuning "REQUIRED" to get INTEL to be compeititve/superior to its competitior....is just too hard to justify this last 5+ years. 9th gen already took out HT unless you got a 9900k....and again...youre spending as much as an entire GPU/SKU "jump up" on a kit of B-die memory...to get the "MOST" out of your intel k chip....meanwhile zen2 was meh with "FLAGSHIP" range gpu...bottlenecking esp in 1-2 core load engines....so ppl had no choice but to go inel....up until zen3 launched....then the choice was 11900k+bdie...the cost/thermal needs etc....vs something like a 5800x cv optimized....and huge $$$ budget towards a larger/faster GPU vs going the intel route talking same BUDGET. NOW in 2024....you also need to add budget towards a HIGH/TOP tier PSU 12v rail/internals etc...for what these powerdraws are looking like with 14th gen. Oh you updated your bios or bios+chipset..and your previous ram/cpu oc/settings are now crashing SHUCKS well lets just act like this isnt COMMONPLACE for anyone pushing their INTEL k chips...that we largely market/release primarily for this purpose. XD
If Intel at least using the baseline profile in their presentation then I can see it kinda okay (but really not okay since they did allow it), but they didn't. For example, the Cinebench that Intel posted shows that 14900K multicore is 1.06 faster than 7950X and from what I see, 7950X scored around 2050 to 2090, thus Intel with 1.06x multiplier should score 2173 to 2215 which it can't achieve without using more than 254W! AMD does use PBO in those benchmark, but from looking at The Verge benchmark, 2058 score can be achieved with the CPU using 224W of power (within AMD spec of max 230W) while their 14900K score is 2194 using 324W of power which is definitely above spec and again, this is the number that Intel use in their own presentation! And no, Intel is not comparing actual baseline (as in purely non OC state, thus 125W) vs AMD actual baseline because if they did, 14900K will lose by a mile. So basically Intel is being hypocritical. Right now if you buy Intel CPU expecting to get the performance shown in Intel's own presentation, you're not getting it with baseline profile. In other words, they are lying about their performance to those costumer that purchased 14900K (or any other K model), running their processor out of spec just to get the win. For Intel to not be lying, they need to allow mainboard partners to default on these unlimited power setting and keep replacing any CPU that run into instability because of it. They can't just say "run the baseline profile if you encounter instability" since with baseline you can't get those numbers from Intel presentation, thus lying.
Push it to the limit Walk along the razor's edge But don't look down just keep your head, or you'll be finished Open up the limit Past the point of no return Reached the top, but still you gotta learn how to keep it
I have a 13900k and didn't know this was an issue until a few weeks ago. I manually tune my system to stay within my comfort zone for silence and under 80C so that might be why I don't have any issues. I'm glad this is being brought to light and hopefully makes intel rethink their view of just pushing more power to make it look good. Hopefully I didn't jinx myself because I said I don't have issues... Would rather not have to deal with a CPU/motherboard swap haha.
Same here. First thing I've done after setting up my 13900k was to limit turbo to 253 Watts manually, I've been running it like that for two months, recently undervolted the core -060 mv as well. Its quiet and trouble-free since then.
@@lenscapes2755 I think I pushed mine hard for a bit seeing it sip 350-400w doing some 6Ghz silliness. But that was pushing the temps up to 90-100C (liquid cooling barely keeping it in check) For daily I it's at 300w PL2, and 256w PL1 and can't recall my undervolt setting. But with that it typically sits around 80C when fully loaded in my hotter than most peoples rooms. (75-80F) This chip really needed the undervolt as that alone lowers the temps by 10C-15C for me. Which in turn lowers the power draw. I'm excited to see how the CPUs of 5 years from now perform. As long as this CPU doesn't crap out on me I don't see me upgrading for at least that long. Maybe even longer if the market gets stagnant like the old days where it was just quad cores for years and year with barely any changes.
I keep the 4096w but use BLCK 100% so clocks only go up to 5.5ghz. Maybe I'm wrong but it seems to me it wont draw nearly as much power this way but is still able to provide enough power to keep stable. I have experimented with 330w limit though but found some instability. I would imagine the locked clock speed keeps it from pushing too hard.
I think intel has known about this issue and just didn't care. Bottom line is no one should have to fiddle with the default BIOS settings for long-term stability. Under-volting for lower temps is fine, but it should not be the go-to for stability in a non-OC situation.
I've always run my 13700k on asus with the multicore enhancements off and a small undervolt, because at defaults it would hit thermal throttle basically instantly under any kind of load. Intel went nuts with the power stuff on these.
@@boy-who-likes-bats Multi-Core Enhancements, It's what Asus calls their default over the top power settings in the BIOS, other motherboard manufacturers have the same thing but presumably by other names. Switching off MCE sets everything to "Intel Stock" power settings but.... as mentioned in all the coverage, what actually is a stock setting is pretty vague. I never dug into exactly what turning off MCS changes, but at the very least it sets PL1 and PL2 to 253w, similar to the "Intel baseline profile" I can't recall seeing what it did to other things, like the svid behavior, iccmax or whatever else. But at the end of the day, all the reports have been around the i9s. I haven't seen any reports of issues with the i7s, and I'd imagine many more of those have been sold, so problems would have higher visibility.
@barrym426 ah, hm. i don't remember if i touched mce when i set my pc up for the first time, but i do remember pl1 and pl2 being 4096 at one point, but i've never seen cpu wattage go over 250 during any load, synthetic or gaming or otherwise. it's been stable and reliable for me for now, i know my chip runs a bit slower than a lot of the oc guys have theirs, but it doesn't get much hotter than 80C, rarely ever. just really not looking forwards to having to buy an entirely new cpu if mine decided to die one day.
Who to blame: 70% Intel, 20% Board Manufacturers, 10% Reviewers for not doing power-normalized benchmarks by default (which motivates Intel+Co to do this stuff)
Can't blame reviewers and that isn't fair if they were included. They do their work with the impression that "stock" means out of the box, pop in and run tests. As a consumer, I appreciate that more than for them to go in and make any adjustments. People who buy these parts are not rushing home to tweak BIOS settings to "power-normalize" anything. Imagine the outcry if reviewers were to make changes prior to their benchmark findings. Then they would be labeled shills for their preferred pc parts to perform better.
@@mazing87 Of course they should still also do benchmarks with stock settings. The issue is that most reviewers only do those. Due to that issues like this linger in the background, and only get real attention when things blow up. The truth is (as can be seen here) that at a cost of 5-10% peak performance Intel could reduce their power consumption and heat dissipation by 20+% (and any user could easily do that by customizing the limits, but very few peole realize that).
Isn't this the same craziness that you were banging on about when you were monitor Steve?! Modern systems man. Go with Intel and you're not running hardware in spec, go with AM5 and your system boots up like a 1980s Compaq or go with NVIDIA and melt your mother flipping 12VHPWR connectors off.
Great video. At work we've been implementing our own version of the baseline limit for almost 2 years. Even the best waterblocks couldn't keep up with these chips when unleashed. Its been frustrating having to set customers expectation on 10-15k water-cooled system they can still see 100c in same situations. Some convert to 7800X3Ds and thank us after the fact. but those who don't some always complain even after we set their expectations. We're a little scare for what's to come with the next gen flagship intel cpu/nvidia gpu combo.
I remember people (including Intel) gave shit to AMD for the 9590 drawing 300w on a 220w TDP. Here’s intel drawing 400w on a 150w tdp on a chip that lives its entire life above 90c stock when playing a cpu intensive game or high refresh rate setups. You can’t even cool a 14900ks to maintain max clock with just a custom loop.
Assuming electromigration is causing the degradation, current is actually the issue, not wattage. A fast corner CPU could run at a lower voltage but higher current and be over the damage limit while under the wattage. Eletromigration doesn’t really happen until you cross the current limit, then it’s exponential.
This is intels fault. You can’t say Intel is innocent because the board makers are boosting beyond defaults - because Intel isn’t enforcing it. They want the higher benchmarks, but blame the mb for the stability. Trying to have their cake and eat it too. It’s same as Nvidia and the power cable. Try to blame the consumer when your faulty designed cable allows for user error. Edit: haha. I wrote this before watching the video and Steve says the exact same thing at 17:22.
It's interesting to see again just how far the power curve these CPUs are being pushed. It was cool to see the 14900K running at less than half power and half the temperature and getting 75% of the performance. I loved that AMD acknowledged it from the get go with Eco Mode for Ryzen 7000.
I initially resolved my issue by downclocking with P-core ratios, but this solution didn't feel ideal. After downloading the latest MSI BIOS with ME 125, I still needed to adjust the P-core ratios. Modifying PL1, PL2, and other settings didn't seem to help. Recalling Intel's statement about a voltage issue, I decided to set the CPU LLC to Intel default, reset the P-cores to auto, and restored all other settings to their previous values. This approach worked successfully. I tested it by running Edge RUclips, Claude, Intel Process Tester, Cinebench, and other applications without any problems. It's worth noting that with these settings, Cinebench R23 scores decreased by about 3500 points compared to when I had set the P-cores to 54 on all but two cores, with those two at 56. However, I would rather have less performance until the voltage problem is addressed. This approach prioritizes system stability and longevity over maximizing benchmark scores.
I really wish at least one motherboard partner would have came out and said we cannot in good faith support the unlocked 13th and 14th gen CPUs. There is no way to build a motherboard and a cooling system without going to excessive extremes to keep these processors safe.
"We're going to be very crisp in our definition of what the difference between in-spec and out-of-spec is." You know what else is going to be very crisp? The CPU, after running at those wattages.
This happened to me the first time I got an intel i9-14900k! It was working perfectly for a while. Then I applied what I thought was a mild overclock to it (nothing crazy, just enabled the AI overclock feature) Than it all went downhill from there. The next day it was blue-screening after launching a game. So I turned off the OC, but it was still blue-screening! I wound up replacing almost every component starting with the RAM first before I finally caved in and returned both the CPU and MLB... I had no idea that these processors are so unstable, just recently I watched Jayz-two-Cents video about the same topic and disabled the default BIOS settings to "enforce all limits"... I only saw about a 3-5% performance hit, but my temps dropped nearly 20 C!
@@Hardwareunboxed similar results with 2 of mine. But I've been running mine at or near 400w for a year with no instability. Running an EVGA KP z690 and Asus strix z690. In fact the only instability I had was going over 7200 xmp on ram. Strix is running 6000xmp and EVGA is running 7200 now and it's solid. I wish we had actual data from these unstable setups because I think there's more to it. Things like ppl running 4 dimms or "Hero" voltage to try and hit 8000m/t on 2 dimms. Sadly the pitchforks are out and the villagers smell blood so we'll likely never get real results.
This would be the real test. With how much outlets love to clickbait this shit you don't know if it's 50 CPUs or 50K. None of the i7s or i9s me or my brothers have built have had stability issues, and I don't personally know anyone that burned up a 7800X3d or 4090 either. Doom and gloom gets clicks and Intel is a juicy target to take shots at.
I wonder how a certain chaser youtuber will react to this situation, unlike a mythical nonexistent problem with amd he made up, this thing is truly can be called Inteldip
@@blegi1245nah, he’s just too ethical to do any such thing. His principles are too high. He’d never lower himself to that sort of behavior. He says this about himself in his videos. We should all believe him when he says things about himself like that that no one asked to hear. 😂
I have a 13900k that I ran perfectly stable (with extensive testing) for over a year. Then suddenly it started crashing a few times a day. Then it became every hour. It wasn't an extreme overclock (in fact, I was undervolting slightly, and running with power limits). I'm pretty sure it's the same issue as whatever is being reported here, because the only way I could resolve it is lowering the clocks. The strange thing in my case is I *did* constrain the power limits to within spec.
In the video he did question whether "in spec" 253W is still killing the processors. My gut says YES, because physics. Until recently, all CPUs were limited to 90-120W TDP. There was no sudden overnight break-through in materials or cooling, so leaping to 200+ W screamed to me that there was a problem.
@@paradoxicalcat7173 I locked my 13600K at 160 watt. You have to be mental to go 250 without delid and custom loop. Now I'm pushing 5.6 P-core 4.4 E-core 4.9 Cache and still fine.
Wow, and some Intel 13900k fans telling me choosing the 7950x3d for lower power consumption doesn't matter... Lower temp, no throttling almost, stable (and now just for 1 year), lower power bill, less expensive cooler required, might last 7 years unlike Intel i9. I'm sad for Intel customers.
I've been watching the 12/13/14th gen users complain for ages about stability, and how 80+ deg. C is "normal" and "OK". AFAIK the laws of physics never changed, and neither did the materials used in the processors, so 80+ deg. C is getting very close to melting stuff. It is now making sense why these things are running so hot and so power-hungry. It's not that they need to be; it's because Intel want performance at literally any price. The price in this case is system stability and greatly reduced component life.
At this point, I really think that going for non-K SKUs seems to be a better bet. I mean, have a little less clock speed, quite a lot less power usage and still good performance.
Intel CPU's are fine all motherboards are set at 4095 watts which needs to be changed nothing to do with Intel ...This has to do with who is building the PC is at fault not intel and i9 13 or 14th gens carries a wattage of 253 watts on the PL1 and PL2 and it should be set at 307 amps for that CPU also for other intel CPU's these numbers would be different u can get the specs of the CPU on intels site unfortunately a lot of people building these systems don't like to read and set the CPU at spec on the boards so don't blame intel people that build the PC's should be blaming themselves cause I don't have any issues with there CPU's cause I set the numbers as per spec.
@@Hardwareunboxed there is no issues I've been running my CPU since it was launched never ever crashed cause I already put the right specs in the board I never go by bios defaults that's not how I build PC's all boards are defaulted to 4095 watts not just 13th or 14th gens all of them as I always check no problems here my friend everyone is making a big thing out of something that people should be checking
Saying your PC is fine therefore it's not an issue is extremely tone deaf. This issue is likely only affecting the poorest quality silicon, higher quality silicon will last much longer, if degradation is the issue.
I'd say, yes, if you leave it the hell alone, a Raptor Lake CPU on Z690 is staggeringly more stable that a Zen 3 platform. My 13600k has literally never BSOD'd or otherwise hard-crashed. Sure, the occasional game will inexplicably crash ( _maybe_ once in three months), but that's just bad coding by the devs.
@@awebuser5914 It seems like the i5s are not really a problem and the 13600k or 14600k is probably the best all around CPU for the socket for the money IMO. I have a z690 and z790 board and will happily run the i5s until I need something else. The i7s and i9s are a problem though and Intel needs to fix it right or they better have a good showing with 15th gen.
I never had any issues with intel CPUs, and i dont have with current 13900k, i dont even know how people manage to push this cpu to its limit in a unreal engine game
I already knew how to adjust the power profiles in the BIOS, but just for giggles I updated the BIOS on my z790 Dark Hero. Originally current limits were flat out unlimited. Intel published new profiles showing 307 amps and 400 amps. When I updated the BIOS, the Intel default current limit was set at 280 amps. This whole mess is all over the map.
@@devans83 I haven't had any issues at 253W with the 400amp Icc max. Any kind of overclocking and I have to walk a very fine line. That said, I don't have any problems running the E-cores at 45X just for a little uplift on the cheap. I'm also running a slight undervolt on the cores and the cache domain. usually -0.15 to -0.20. Your mileage may vary.
Thank you for diving deep into such issues, showing what is actually causing this whole issue. Seems like the only way to fix this problem is reengineering the CPU core architecture in order to achieve high performance at a relatively lower power level...
I am using a Soyo (basically a rebranded Maxsun) AMD B550 board. Other than the initial pain with the BIOS (am not sure who's at fault here, me or Soyo), it's been running excellently with great temps and stability. Decided to try out Soyo because of good memories of the original Soyo back in the 90s and also good reviews from youtubers about the board.
@@DonaldLim-r2t I never heard of either of them until now, from what I can tell they both must be fully chinese brands because they seem to only be available from aliexpress, but both are worth to look out for.
This happened to me as well. I built a new PC in January and my first combination was an MSI Z790 Tomahawk paired with an Intel 14900k CPU. After about a week or so, started getting random crashes which got worse to the point where i would get the out of memory error when booting up games, such as Hogwarts. I did a lot of troubleshooting, which I will spare the details here, and ended up switching motherboards to the Asus Strix Z790 Gaming. This was not the issue. I ended up replacing the CPU to another 14900K and this worked immediately, however, the problem came back after 1 week of use... I was lucky that I was within my stores return policy. That said, I couldn't trust Intel any more and swapped everything out and built a new system around an AMD 7800X3D chip, which has been stable ever since, albeit not as fast. I thought it was a bad batch of CPU's, but now seeing these reports, it sounds like this issue runs deeper. Seriously, what is the point of buying a premium motherboard if you can't take advantage of it pushing the CPU???
You completely and utterly mis-understand the problem. The problem is NOT the motherboard. The problem is Intel running their CPUs way beyond what the silicon can handle.
@@paradoxicalcat7173 I'm pretty sure I was clear in my comments that it was not a motherboard issue. I was simply stating that during my troubleshooting, I ended up swapping motherboards since I couldn't figure out the source of the "out of video memory" problem. Having built many systems in the past, this was my first time experiencing this type of degradation in a CPU and didn't think it was possible without a hard overclock. I ended up trying 2 x brand new 14900K CPUs and both of them failed after I installed them in 2 higher end motherboards, running at their default settings with only XMP enabled. Intel has confirmed in the past that this is not "out of spec" to have this enabled. My last comment regarding the point of buying a higher end motherboard as being useless, is because these motherboards push the CPUs to the limit, which obviously causes some to degrade prematurely... If Intel now points to these motherboard manufaturers as being the problem, and need them to remove any type of "boost", then you might as well buy a basic motherboard. That's all I was trying to say...
And they have a bunch of them. 13900KS, 14900K, 14900KS, are just bins on top of the top bin. For this, you get the privilege of supporting the company with your money.
Remember when auto-OC became standard and people all over the internet were concerned and questioned if this was safe for their CPUs and their concerns were dismissed because of all the built-in safety features and even manual OC was dismissed as unsafe compared to auto-OC?
Oooh cheeky wee sunday morning one. Its like when GN listened to ASUS and sarcastically ran the AM5 socket completely stock everything and showed how performance was lowered 🤷
As someone with a 13900k and MSi board I can say it works. You get similar performance to the ASUS profile shown here for 253W and it dials back the voltage and heat quite a lot. If it's at 4096 it will push voltage above 1.5V and will hit 100C. At 253W it's a 10-15% drop in performance, hovers around 1.38V and it runs at like 80-85C. I think MSI bios acts different to ASUS and Gigabyte but I could be wrong.
They clearly did mentioned another step to select “intel default” under CPU lite load control option. But this will use higher voltage compared to MSI’s default option. Not sure whether it’s better to switch to intel default for this step..
17:50 Given the power use and process node it seemed possible to me Intel was trying to max their chips out to compete with AMD at 5nm. Seems very likely now. Not cool to given high spec CPU users are now paying for it.
How is motherboard BIOS settings that push a CPU past its intended operating limits ,which results in the CPU crashing, Intel's fault? My 14900KS is perfectly stable, and has been from day one... BUT, it depends entirely on what revision of BIOS your using. Im running at 8000mhz 24hour stable memory speeds and the CPU is functioning with zero stability issues.
It's Intels fault mainly because the intel baseline is not usually the default setting out of the box. Intel made the CPU so they should be enforcing the baseline power limit as default. Any changes to this by the user is then the fault of user if the CPU degrades. AMD just had a similar issue with 7000 series CPUs frying because they didn't make board partners enforce a voltage limit and let the board partners push things way too far. Intel and AMD really must watch these board partners from now on to make sure they don't push things way out of spec because it's bitten both of them in the ass.
The motherboard manufacturers have to sign a contract with Intel to get chipsets and firmware to build motherboards. So Intel has all control in the world over this. But they want to write "TDP: 253W" on the box of the CPU (so it doesn't look too bad) and then have the board blast the CPU (with plausible deniability) to win benchmarks. So yes, it's Intels fault. Very simple. Nvidia doesn't allow their partners to just do whatever with their GPUs (much to the frustration of EVGA, but that's a story for another day). To get an OC VBIOS signed by Nvidia (otherwise the OS driver won't talk to it) the board partner has to send Nvidia the VRAM layout and component choice along with a standardized performance test of the cooler. Intel has the power to do this too. But they chose to make it very opaque what who is actually defining what "TDP: 253W" means in any particular case. For reference: Not a single professional desktop machine from HP, Lenovo or Dell runs Intel processors differently from Intels "guidance values" (And trust me, I've checked a fair few). Because corporate decision isn't made on benchmarks in popular media. In this environment power consumption actually matter, and 5-10% performance difference gets easily washed out in service contract details. So here Intel are magically honest and proper professional. Who would have thunked!
One of the big problems of relying on online reports of crashing games it that games often have serious bugs anyway and it's hard to distinguish between the game crashing because of it's own problems or the underlying hardware.
This is true, however there's a lot more to go on here with multiple companies conducting investigations and all pointed at K-SKU 13th and 14th gen CPUs being the problem.
@@Hardwareunboxedmy AVX 512 enabled i7 12700 non k with asus tuf z690 plus wifi d4 at the cost of single i9 12900k in india is still going well since launch jan 2022 😊
@@predabot__6778And it's K-series CPUs that suffer from this, since those are overclockable and the motherboard basically overclocks them out of the box.
My i9 13900KF, which is now over a year old, was a champion at the beginning, without any issues. However, now, without enabling the Intel Baseline profile on my motherboard, I crash quite frequently whenever I play Remnant 2. I faced this same issue previously as well, but it's worse now. The Baseline profile fixed this, but at the cost of performance. *Take this with a grain of salt. I overclocked my CPU a few times, using my motherboard profiles and a bit myself with the help of professional RUclips videos. Though in my humble opinion, it didn't get worse, but in benchmarks, it would be unstable, even without the OC profile. However, a simple reset of my PC fixed that.*
If your pc continues to crash in benchmarks you need to adjust your motherboard's LLC calibration because your CPU voltage drops too much under higher load. find out which profile is stable and doesn't overheat your CPU
@@MiGujack3 7600x is still much better for gaming then i9 9900(nok K) and i just wanted to switch to am5 with affordable option atm. Next upgrade will be zen5 9000 series AND i don't have to change my board. No real brainer for me for sure.
I have this problem for almost 3 years ago for intel CPUs in my workshop from 12th gen to 14th gen because it consumes a lot of power and my customers complain about the heat it goes up to 95 degrees in any 360mm aio or 420mm liquid cooling system 😢 and i decide to lower the power consumption of the CPU or limit the overheat temperature protection to 80 degrees and that's works for me some customers doesn't care about the frequency of the cpu they care about the heat and some other care's ...I tell them the performance will be less about 7 % and they are satisfied for that ....everyone is happy ...they happy 😁 me happy ....for every build i build it for intel CPUs and I believe the common enemy for electronics it's the heat More power = more heat More issues..unstable.. crashes ...for me it's like overclocking unless you have the best power stages or best VRM components and best cooling system to gain this much power
Intel cannot command motherboard makers do anything but in the Intel white papers you can see the power limits. Page 13:46 in video. This is why both Intel and AMD CPUs should be benchmarked at stock, with stock RAM. Ask yourself which companies dont have to change their benchmarks? The ones that do have to change their benchmarks because of this are the ones that dont know how to benchmark correctly. AMD CPUs are massively overclocked in benchmarks as well. AMD CPUs were burning IO dies to death because the motherboards were putting way too much voltage into them. Yet benchmarking sites cant read the documentation and do stock benchmarks. So we have extreme overclocking on both Intel and AMD as standard. DDR5-6000 RAM etc to make AMD cpus faster, so you get the maximum out of their IF @ 2000 MHz. More or less the maximum you get out of AMD. This is massive overclocking as well. Reviews are the problem, their benchmarks are all massive overclocks. Also Intel they dont increase the cache frequency like they do with AMD's IF. AMD's IF is maxed out. Both together on both CPUs equal a big bump in performance. I blame sites/channels like Hardware Unboxed. Its their biased benchmarks that lead to motherboard manufactures maxing out everything. Stop overclocking your Cpus in your benchmarks and watch the problem go away. Stop blaming Intel because you dont follow their documentation. Look at the state of the AMD overclocks in your reviews. How is that fair, their IO dies failed due to the voltage of getting those IF@2000 and DDR5-6000 overclocks which you used in all your reviews. Also stop pretending to be experts, your not. If you were you wouldnt be rerunning to see the performance loss. Dont act like its the fault of someone else as experts you would have limited the power draw of Intel CPUs and not overclocked AMD CPUs to death. Increasing IF to 2000 and using DDR5-6000. Really biased for AMD stuff. Hopefully AMD payed you for your reviews with free hardware. Maybe they told you how to setup the system. If you can find a benchmarking website that didnt have to change their benchmarks. Then you know they are unbiased. There is an overclocking channel laughing at all this, because they could read Intel documents and saw this comming. Same with AMD's IO chip getting way too much voltage. For us overclockers, we know you cant trust benchmarks. You know results close to your maximum overclock on a silcon lottery winner cpu. AMD CPUs are by far the worst for special tuning before benchmarks. From Intels website Intel® Core™ i9 processor 14900K : Processor Base Power 125 W Maximum Turbo Power 253 W Memory Types Up to DDR5 5600 MT/s Up to DDR4 3200 MT/s Were your benchmarks running these power limits and memory speeds? From amd 7800x3d Default TDP 120W Max Memory Speed 2x1R DDR5-5200 2x2R DDR5-5200 4x1R DDR5-3600 4x2R DDR5-3600 Were your benchmarks running these power limits and memory speeds? If not the problem is your benchmarking.
All the power specs are in the datasheet, volume 1 of 2, including Extreme Config. At no point in the datasheets does it say “no limits.” Unless Intel writes it down somewhere for the public to see, they’re covered. The “baseline profile” varies from CPU to CPU. It is not a one size fits all setting. Gigabyte is using the 6p+8e 125w CPU settings instead of setting the values for an 8p+16e 125w part. This isn’t rocket science, and MSI has handled it for years. On a new CPU install, MSI asks what cooling is being used - box, air, or water. The board then sets the power limits accordingly.
I had a MSI build and it defaulted to "water cooling" with all the limits removed. In fact none of them are Intel specs, both tower and water cooling are both unlimited at 4096 Watts with only box having a sensible PL2, but PL1 is set too low at 65 Watts even for unlocked CPUs
Most of my friends on Intel are running the i5 12400 and haven't upgraded. No E-cores so no need to use the application optimization software, and no crazy power consumption.
188W looks like halfway between 125 and 253 (minus 1 watt) and thus like a compromise. There is no other SKU that has a PL2 as default, so not likely an error recognizing the chip or something like that. There used to be a formula for PL2, like PL2 = 1.25 x PL1. That was too constrained for the higher core count CPUs so now it adds a 'nice number' (128) to the 125W 🙂 I understand Gigabyte for playing it safe here...
I was looking for one of those recently. In New Zealand people pay about $400 NZD used, while the 12400F is $240 NZD new, and it out preforms, with much lower temps.
Wait so what about those of us who enjoy gaming but aren't as "in the know" in regards to this stuff? I've got an i9 14900k and an asus motherboard, while I haven't had any issues is there anything manual I need to do in order to future proof?
Gotta love how the "intel baseline" isn't the same across motherboard vendors. Almost like the intel baseline isn't provided by intel but cooked up by each vendor on their own.
Also it's worth noting that SVID set to intel fail safe runs more Vcore than the ASUS default settings for any scenario that doesn't hit the power limit. Gigabyte's intel baseline profile does the same. Most gigabyte boards use AC_LL and DC_LL of less than 1.1mOhms on default but with the baseline profile the AC and DC_LL get set to 1.7mOhm which raise the core voltage for any workload that doesn't hit power limits.
It's pretty funny really, true to form :D
It almost feels like these "Intel baselines" were something nobody cared about for years, so much so, that now mb manufacturers are really scratching their heads.
"Wait, what, there are Intel baselines? Who says?"
"Dude, Intel, of course, who else. They're blaming us for not abiding by them."
"Oh-oh. Ok, let's fix it at once then. Give me the numbers."
"Yeah, about that... Thing is, we are not sure we have the numbers. Trying to contact Intel to provide us with them, but they are keeping us on hold for several days. The last guy said he'll go to look in their archives and the one before that said he'll consult the engineers and we haven't heard from either of them since". 😁
The part that bugs me is that these kind of thing have been a rule and not an exception for a while and there seems that there haven't been blatant issues up until now. This bugs me because it narrows down the potential issues to be likely something along the line of a bad batch or suppliers of materials or even worse a known design flaw somewhere within the chip. I can also get why they would be cagey about it because of the possibility of some kind of class action involved as well. If you tell people they are buying "top of the line" type parts and then they end up with the same performance as a middle of the range I7 (Don't quote me on that, I haven't checked it up again) due to how power limits magically need to be enforced now that is going to make a lot Intel buyers start frothing at the mouth. Also it means we should probably rename the Intel baseline specs to "should-have-bought-AMD-mode". I know I am being really cynical and sarcastic but this is going to have serious consequences for some people.
Lasty , dumb question has been bugging me. Has anyone seen any positive or negative correlation between cooler power and the issues with these chips? I cant figure out if having better or worse temperatures would mean more wear on the CPU in this situation. Is lower temperature but higher voltage equal or worse than higher temperature lower voltage and throttling for these CPUs?
@@Hardwareunboxed
Will laptops have the same problems?
Like what you have told today ???
also the way they responded to igors by using the word "reccomend" not "must" is just horrendous
Their response looked like "if your processor or Motherboard isnt dead yet then it is in spec" lol
Funny when i put defaults on motherboard msi carbon z790 thats when windows crashed however when i put a balanced OC no crashes since didnt mess w the long / short duration should i be worried? Temps are in check too
@@adamtajhassam9188I’m confused how you ended up on z790 in the first place. Just cont let go of team blue I take it?
@@shiraz1736 Why not? My 14700k is a very strong performer across all uses. And in gaming I like the higher minimum 1%/0.1% FPS I get vs. AMD on the same games.
@@shiraz1736I bough intel even knowing that thers problem long time ago with motherboards overclocking cpu in default bios settings . It's strange to not check all the info on internet before buying. And on asus motherboard is only one option to click in bios to change it. And still for me intel won. I was choosing between 7950x3d and I 13900 k and Intel was cheaper, same motherboard was cheaper and I know it that any ram I will buy will work. And I know that on amd only 8 cores have cash and I didn't want to play with core scheduling for every application that I can use.
@@B.D.E. Ok that 1% is a seller for sure.
Intel: Motherboard manufacturers, please make sure we win in benchmarks
Also Intel: Motherboard manufacturers are causing the crashes on your systems by running out of spec
Intel: You know that spec we said was 'in-spec' for the past 5+ years?
Board Makers: Yes
Intel: Well it's out of spec now dummies, FIX IT!
Board partners: Out of spec? What is out of spec?
Intel: Out of spec is in spec is out of spec.
Board Partners and reviewers: Wut???
@@johnd.1618 Stop you are not helping AMD or its users with this kind of comments, acting like a fanboy is irrational, neither AMD or Intel are ours friends they want our hard won money and that's it. And if you are wondering, I'm a AMD Ryzen user.
@@johnd.1618just because most people don’t think on as low as a level as you do doesn’t mean we dint understand. We just understand so much more than you could. Ryzen user for years here.
@@johnd.1618Steve specifically mentions they've been bringing up this in-spec/out-of-spec nonsense for years, even prior to 13th and 14th gen. It's only now that power delivery's gotten so out of hand that the chickens have come home to roost and at an accelerated pace.
HUB and other tech outlets have been mentioning the writing on the wall for well over two years. What're you talking about?
Userbenchmark declares war on Hardware Unboxed, will send a pre-emptive strike of unhinged text on their website.
Ohh nooooo :D haha
😂😂😂. Their large user base will be throwing party that Intel baseline profile pushed efficiency better than 7800x3d,7950x3d
@@vasudevmenon2496 Their dealer will be like "you sure? That's 1.5x your usual dose:"
@@kloroformd haha
@@kloroformd more copium please.
Waiting for Userbenchmark to explain this is because of amd's marketing.
It IS because of AMD marketing though..........
"AMD cheated by making CPUs too efficient, forcing Intel processor to cosplay as a nuclear reactor in SSR of Ukraine!" - UserBenchmark, probably...
UserBenchmark: "The Intel Core 14900K is a 24 Core processor, and the fastest CPU in the world. It has a massive 6GHz boost clock, which is much higher than the Ryzen 7950X3D's boost clock of 5.7GHz. Gamers or Creators will have very little reason to look at the 7950X3D or the 7950X due to it costing $100 and performing roughly 10% in single-core and multi-core. However, an army of AMD's idiot baboon fanboys like Gamer's Nexus and Hardware unboxed will still try to convince you that the 7950X3D is actually worth it, and claim that Intel's 14900K experiences 'stability issues', even though that is only typically the result of extreme high overclocks. Even with the baseline profile, the 14900K beats the 7950X3D in single-core, which further proves the worthlessness of AMD's processor, and the fact that Ryzen CPUs are no better than Bulldozer CPUs. An average user or gamer should not look at any of AMD's processors, as doing so will only result in disappointment due to lower performance, more instability and worse software."
@@christophermullins7163 How ? did they like market that we're proud of CPUs exploding and then Intel got into FOMO mode and said we gotta burn our CPUs too ? lol
@@lucidnonsense942 Nah cosplay as a rocket engine 💀
The best proof that power limits on Intel are a horrible mess is that even 'Intel baseline' profiles are different. It's great that - as usually - you try to get to the bottom of it instead of believing Intel marketing trying to deflect blame.
It's pretty clear what the baselines are, Steve already showed it in the graph. Never try to make sense of what Gigabyte does.
Intel specs on white paper are really clear, maybe reading and comprehension is no more a request.
@@johnd.1618 Gigabyte sure put much bigger breaks on performance than everyone else.
Overclocking used to be a deliberate choice - one for people willing to risk stability and deal with higher power, for marginal gains. But these days, out-of-factory overclocking seems to be the new default, and you need to deliberately go and limit everything yourself to get back a cool and stable system which performs pretty much the same...
It wasn't always marginal
It is now marginal. Before it was risking stability & higher power for noticeably higher performance.
In the past it wasn't rare for manufacturers to play so safe with default base performance that it was feasible to get about 10% performance increase without doing much to efficiency. GPUs such as the HD 7950 allowed pretty much every owner to increase performance by a higher % than you can get nowadays changing a 4060 for a 4060ti
@@Rentta I guess it depends how recent is "recent", and how marginal is "marginal". Something like 2600k was more than ten years ago. What's the most recent mainstream component that was still decently overclockable?
@@RenttaOverclocking was usually in the neighbourhood of 10% gains, which IMO is somewhat marginal. Is 66fps that good if previously you had 60fps? I'd wager that no, not really. You need significantly higher performance if you want to feel a difference.
Remember the old days when processors had so much overclocking headroom or you could even re-enable disabled cores in some cases? Now they run them right at the edge of stability, or in Intel's case not even that
Ah Yes, phenom x2 enabled to x4 and OCed to match the performance of 4x more expensive cpu
You can still enable and disable cores relatively easy, with windows and CPU vendors tools, in case of AMD with Ryzen master app for example.
But this can also be a problem, because people forget stuff, I know one person that disabled cores, then forgot about it, latter made a angry youtube video saying that Microsoft windows was dumb, because "it didn't saw all the cores of his CPU!", when in reality like I said he manually disabled the cores and just forgot...
@@Argoon1981 true, but I'm talking about old 3 or 2 core processors that were sold that way but actually had 4 usable cores in some cases. I think there were even 4 core Phenom IIs that could be unlocked to 6-core
i swear the new amd chips come out see 95c thermal profile...ppl gasped...etc....then quickly realized it could be changed to 75c....and still equate GREAT workload performance and equally great in gaming/shared loads. What a circle back in usage.....meanwhile trying to uv a "fresh and new" intel or amd laptop on the otherhand....those things are rigged to blow and marketed to be "upgraded" on a constant basis to their PURE 'laptop user" demographic its a sad state of affairs. Granted huge leaps in performance on the mobile front in the last 5 years...in realworld uses and theoreticals. The overall thermals/powerprofile aspects still has me largely thinking this gen is worse from a stability standpoint and system health over the course of ownership. And ive built unlocked phenom x2 DFI lanparty rigs in the past...even hybrid crossfire rigs with APU and overclocked chipsets....weird how theres more headscratching involved and general system/degredation in play these days if you want to tune a system or PUSH a budget system to its limits(in the past) considering you cant even really do somthing on par with a "buffed" budget build anymore.
Capitalist innovation. Except innovation comes from how to extract more money, not to make a better product and make money because it’s better.
it stinks of desperation by intel. When you sacrifice stability for performance you start going down a slippery slope.
FX bulldozer flashbacks. I hope we're not heading towards that.
That's all the 13th and 14th gen CPUs are is just more power hungry versions of 12th gen for the most part.
You can be absolutely sure the OEMs like HP, Lenovo and Dell are being told: "You better fckn set the values we 'recommend' on your professional lines! What you do on the consumer crap is none of our business, but if the professional lines aren't stable we will crucify you with a jackhammer!".
@@DragonOfTheMortalKombat Oh they are fully going towards that. Intel had a clever idea of putting in smaller E cores for power efficacy like mobile phones had and while it's a great idea for laptop when you pair 8 E core with 2 P cores. It's basically worthless for desktop when even after windows added support for it the E cores largely don't do much and because of the whole E cores idea the P cores (while powerful) are super inefficient. Just like how bulldozer had a great idea of boosting multithreading at the expense of single threading. Intel is in the middle of needed to double down as a architecture is designed to last a decade or more phase and you can't just turn the ship that quickly. So they're dumping tons of power in it to compensate. I really like the big little idea but for Intel the big cores just lose to Zen and the little cores aren't good enough to replace a multithread so they aren't really needed.
a sloppy slope
“I used to be in spec, but then they changed what in spec was. Now what I’m with isn’t in spec anymore and what’s in spec seems weird and scary. It’ll happen to you!”
-Intel, or motherboard manufacturers. Or Grandpa Simpson.
"in-spec" leads to severe silicon degradation. thank you intel!!
The less they live the more CPU's you sale...
I don't know if it's silicon, because we know there is heatsink contact problems with the 1700 pin stuff, so I would be more inclined to say it's thermal paste degradation.
@@SidneyCritic
truth is we don't know exactly. but if thermals were a problem, wouldn't the cpu just lower the clocks instead of crashing? crashing means the load, clocks and voltages are no (longer) ok.
one (of many, granted) probability is silicon degradation.
@@SidneyCritic It's the silicon. Degradation causes the CPU to run a workload with instability, forcing you to undervolt or change its clock speed to continue working stably. It's the same thing that happens if you overvolt a CPU and degrade it, you can no longer run it at the same clock speeds and settings without it being unstable.
@@riven4121I guess you meant overvolt instead of undervolt? After degradation takes place, the CPU will require a higher voltage to mantain the same clock speed as before.
Agree. AMD did a really good job for the 7800X3D disaster.
They reacted to the burnt issue very quickly, admitted the mistake and presented the fix before it's getting worse.
Now, no 7800X3D will burn and it's still seating in the crown.
While Intel just keeps finding excuses 🤐 and being bro with Userbenchmark.
I don't think userbenchmark has anything to do with Intel and also AMD wasn't really at fault for exploding CPUs, mobo manufacturers were with their insanely high memory control voltages.
@@johnd.1618 That's not the point. The point is that even Intel doesn't really approve of userbenchmark. That side is run by someone who most likely has mental issues.
@CanIHasThisName You misspelled "a large Intel stock portfolio".
@@CanIHasThisName How do you know that Intel doesn't aprove of that "reviews" site? Have they made a public disclaimer discrediting it?
When Mobo vendors did ignoore AMDs set fail safe guidelines and disabled fail safes, AMD shut them down by enforcing it with a new firmware. as intel should have done years ago.
Kept going back and forth with my new build. Ended up switching to AMD for the first time in about 15 years. Decided to get a Ryzen 7800 X3D, with a RTX 4080 super.
Same
Hope you enjoy 7800X3D the legendary gaming chip
Don't forget to check out how to use AMD's Curve Optimiser in the BIOS to moderately undervolt your CPU and get extra performance out of it. I'm running my 7800X3D at -27. Some people can do even better but beware benchmarks that rely only on Cinebench as that's small enough to fit in the L3 cache and doesn't stress the memory subsystem. Even so, the irony that not only are AMD CPUs far more efficient than Intel's to begin with but that you can tweak them to be even more efficient to get more performance versus Intel's approach of throwing voltage at everything is too delicious for words.
@@TheRealPotoroo thanks for the advice.
@@SOMEONE23145 Thanks, that's exactly what I was thinking as well. I went with the 7800, thinking I could upgrade in a year or two. I got the MSI X670E GAMING PLUS
The 'where can I Blu-Tac a processor to for B-Roll?' game is pushing the same level of your continuing effort to highlight anti-consumer practice and I remain here for it!
I continue to be thankful that I went with the 7800x3D despite the rocky start.
Hey, how much improved AM5 until today? I'm currently planning my first build but I'm unsure what Platform to pick
@@BigFoot47-48 Mine's been pretty stable for several months, so I've been happy with it. The thing performs well in games and is easy to keep cool.
@@BigFoot47-48I've had a 7800X3D since September of last year. I've had no issues. I concur with Josh. It's easy to keep cool. I have an Deep Cool AK620 Air Cooler. The upgrade from my 5800X was substantial.
@@BigFoot47-48 Run hardware info 64 and run your games & watch vsoc like a hawk....I undervolted pbo disabled eco mode the works & helldivers 2 would spike to 2.2v and hit 141 c ! granted it was for a split second but the repeatability promted a return. newest bios can help it depends and varies.
I feel the same about my 5800X3D and RTX 4080 build, as I've had absolutely zero issues and the gaming performance has been fantastic.
9:25 I can confirm that this has to be a bug in the Gigabyte BIOS. I updated to the latest BIOS and saw the same issue (not with an i9, though). Boost behaviour was completely broken and especially the E-cores didn't boost *at all* for any period of time. I saw a performance loss of ~30% with these settings. After reverting back to the previous BIOS version and loading my own baseline profile (where I basically only set the power- and current limits to the "recommended" Intel values), everything went back to normal and for some odd reason performance even increased slightly, while temps and power draw were fine.
Well let's hope it's a bug...
I also saw a performance boost when I undervolted my i7. It went faster and stayed cooler without approaching 100c in cinebench anymore.
Jay2Cents showed this behaviour in a recent videos.
@@zodwraith5745 The reason under-volting increases performance is because it is no longer thermal throttling. Thermal throttling isn't binary like the old days; the speed ramp is relative to delta temperature and rate of temperature rise above the threshold.
I also have an i7 and under-volted it from stock to 1.2V. It OC no problems to 5 GHz on air, and has run this way for 7 years.
@@paradoxicalcat7173 I didn't say it wasn't. I just didn't bother explaining _why_ undervolting is helpful.
SI here, we sell prebuilds with i9-13900K and I9-14900K. Mid range models with b760 boards that stick to Intel stock settings. And two 14900K failed with us was on stock limits. I have video proof as we were baffled that the degradation happened at stock power. We have had 13900k fail on b760.
if u using bios default, it have OC already.... u have to enforce all limit in BIOS setting not using default
@josephlai99 As shared above, i am working for an SI. So bios is our main game. In case you didn't notice, i am talking about a b760 chipset board that doesn't allow overclocking. Yes, brands like Asus enable Asus performance enhancement by default on all Asus boards. But this is disabled in the midrange options. Every system is stress tested and benchmarked while measuring power and temperature as part of our quality control process. So 253 watts, that's all that goes in under any condition.
Ofcourse we have higher end models using rog strix b760 and z790. These boards with asus performance enhancement enabled pulls upto 390watts peak. We had failures in stock power.
I've even had a 13500 fail on a MSI B760 board.
Edit: Turns out it's a RAM issue, a simple replug fixed it
Really...you should never trust "default" or "stock" settings, ever. I have ALWAYS checked the defaults to ensure they are reasonable. A power limit of 4 kW is NOT reasonable. It would raise a huge red flag for me, and cause me to set the power limits to whatever the spec sheet defined. 253W is a very high figure still, given that until now, most CPU packages have been limited to 90-120W due to thermals. YOU CAN'T CHANGE THE LAWS OF PHYSICS, but you can sacrifice your device for short-term performance.
It's like drag racing: you can take a 5.7L V8 and make it produce 4000 HP, but it can only run 1/4 mile before it needs a total rebuild.
I am guessing those 760 boards are also strictly entry level, including ability to run a cpu
Jufes from Frame Chasers has been really quiet since this situation had come to light 😂
Too busy scamming his audience 600 bucks for unstable intel memory overclocks.
@@blegi1245😂 exactly!
he literally picked up where Silicon Lottery left off
Someone who OC's won't be having these issues, so no. They don't care as it won't effect them or their customers
Idk what I been running a 13900k all core oc 5.7ghz none of those issue passed prime95 48hrs etc
FrameChasers has been real quiet
Stability is not his safe word.
His fans would just buy another 16 CPUs to find one capable of running the speed they want. Each to their own I guess.
Too busy scamming 600 bucks from people for unstable intel memory overclocks.
I think he had to admit 7800x3d is the better gaming CPU some time ago even with all that min-maxing of Intel CPUs and die cooling, and internal hate for anything with AMD on it (not sure what his problem is).
@@andreiga76 Well we have also never figured out what Userbenchmarks' problem is. Maybe they both don't like a female CEO or the colour red ? who knows lmao
So it was “in spec” when performance was needed for benchmarks at launch, but it’s out of spec when consumer wants said performance. Noted.
Its pretty shady that Intel threw the motherboard manufacturers under the bus in their statement. They allowed them to do that the entire time (and may have even encouraged it).
They indeed _have_ been encouraging it or at least conveniently overlooked it for more than half of the last decade. _Since at the end of the day, the _*_only_*_ lone beneficiary is Intel itself_ …
I mean, didn't Asus engineered their M.C.E in collaboration with Intel itself?!
*They deliberately skewed benchmarks ever since and freely allowed reviewers to run **-out of spec-** **_in spec_** and thus, glossed over it as being totally fine.*
Meanwhile reviewers (pretty much all of them!) shadily were happily hiding behind the alleged delivery-condition and every question regarding power-draw, heat-dissipation and whatnot, was disregarded as running just the usual 'out of the box' experience, when in fact *_everyone_*_ involved in this mysterious Bermuda triangle of Intel, OEMs and reviewers _*_knew very well, that *exactly nothing was ever actually running at official specifications_* ...
Yet no-one dared to care but held open their hands for their share of hush-money (OEMs got their rebates, while reviewers are paid in cost-free samples).
All this is done since ages on the back of the clueless customer and his pockets of quicker dying parts and his power-bill, while never ever going to get what he actually paid for, since it's pretty much non-existing. Since if Intel's parts were running at actual specs, Intel-CPUs would rate a very poor second, while being still power-hungry, inefficient and a hot-mess.
The best part of this is Intel was happy with the increased power limits on Publications because it gave them good performance in reviews on release. They've been doing this for years and it finally caught up to them. AMD is doing something similar with PBO and increased power limits but not nearly as severe. All these extra default boosts on motherboards are why people can't get much of an overclock because it's running beyond the default specs out of the box.
AMD is much stricter here and have become increasingly so since the AM5 issues, for example PBO cannot be enabled by default.
It wasn't an issue with those older CPU's because the voltage stayed within the higher end of reason. Now the VCore spikes up +1.5v on these processors.
The old rule of thumb for a 24/7 overclock was to keep the VCore below 1.35v. Intel has deluded themselves, vendors, and noobies into thinking 1.45v to 1.55v is safe, when it isn't and never was for long term stability.
There is a huge difference. AMD does not consider PBO as 'in spec' or 'stock'
So long story short, without those baseline limiters to keep the CPU healthy and adjusting for silicon lottery, AMD's cpus are technically faster than Intels all these years. AMD cpus run cooler and at half the power meaning that they have headroom for overclocking
@@m8x425 That's not necessarily true, it depends on silicon quality, masking quality, and how well the processors are designed internally to regulate voltage and prevent fall-off.
Over time, you would generally expect safe v-core voltages to increase, particularly as processors get better at having consistent voltage across all transistors.
Can't wait for Zen 5
11:55 don't think we did not catch that joke
Please help me. I don't get it :)
@@tobytoxdIt's about bad ole' Palps...
@@LupusAries Wow, you are helping so much, when you speak 3 words in slang. I googled your words and also specifically just "Palps". Didn't know, i wanted to learn more about insects.
@@tobytoxd;) I didn't want to spoil it by saying it was about "creamy Sheev" Palpatine/Imperator Palpatine.....there is a rather infamous scene where he goes "UNLIMITED POWAAAH!!!" ;)
@@LupusAries Ah, thanks! That's ensaberlightening to say the least. As a German i have to admit, i did watch those movies in that germanic accent, which didn't end up as famous as the original. SHAME! .)
This is why I like Hardware Unboxed; they always remember. I saw a few outlets and youtubers making this a "new problem", when I remember my Z170 ASUS Maximus VIII Hero (Skylake CPU) shipping with MCE enabled by default already.
//edited to correct Intel Chipset (Z170)
I can also recall plenty of reviews and other videos where GN Steve ranted about this
same as Asus on Z690.....bios default set to AUTO... MCE ON already
PL2 of 4096watts?
god damn i knew i needed a nuclear power plant
It just means no limits.
not really, it just allows for further overclocking.
Relax, it’s a bogus number. No Intel CPU uses 4096 watts, obviously. 😅
@@disco.volantemaybe not, but maybe motherboards pushed CPUs a little too far when the power limit was removed and when the CPU needs more voltage for a given frequency, it just gives up, can't handle that and there you go, instability
@@prosecanlik4296 Given how high the frequencies were, Intel was going to run into this problem sooner or later. Intel 7 node was initally made for the 5.2 Ghz 12900K. That some node has been used for a 6.2Ghz processor. Any good overclocker will tell you that this much increase in clockspeed is simply not sustainable for everyday use. It is way too unstable unless you increase voltage a lot which is harmful to the processor. On top of that, all these CPUs will happily go to 100C before downclocking or throttling.
Why?
"longer bar better!"
IDK mate, the one's to blame is surely AMD. AMD's parts being so power efficient made intel pursue the same performance without the efficiently, causing all this stability issue. Without AMD we would still enjoy super stable intel CPUs, which might be a little slower and lack some half the cores.
...what did i just read
@@auritro3903a wealth of very gentle satire
@@kenshirogenjuro873 ah, touché
And 4 core ones at that...
and of course a 32 bit only, no any 64 bullcrap.
This fiasco has certainly reduced my expectations from arrow lake now. They are definitely in HUGE trouble.
Since 12 gen they are just cranking up the power to get a perception of better performance. Since 13th gen i thought that it couldn't be infinity and the next architecture should have problemas with it (14th gen uses the same).
Yes I am curious how the next gen if CPU will fare. They are already removing hyperthreading to get more performance and should use lower power by removing some instructions. However will that be emough?
@@pedro4205 the 11900k was a crackhead too
@@m8x425 Yes, but a single generation doesn't show a pattern, And by that time it was still under 200W
the sheer amount of tinkering/work b-die memory/tuning "REQUIRED" to get INTEL to be compeititve/superior to its competitior....is just too hard to justify this last 5+ years. 9th gen already took out HT unless you got a 9900k....and again...youre spending as much as an entire GPU/SKU "jump up" on a kit of B-die memory...to get the "MOST" out of your intel k chip....meanwhile zen2 was meh with "FLAGSHIP" range gpu...bottlenecking esp in 1-2 core load engines....so ppl had no choice but to go inel....up until zen3 launched....then the choice was 11900k+bdie...the cost/thermal needs etc....vs something like a 5800x cv optimized....and huge $$$ budget towards a larger/faster GPU vs going the intel route talking same BUDGET. NOW in 2024....you also need to add budget towards a HIGH/TOP tier PSU 12v rail/internals etc...for what these powerdraws are looking like with 14th gen. Oh you updated your bios or bios+chipset..and your previous ram/cpu oc/settings are now crashing SHUCKS well lets just act like this isnt COMMONPLACE for anyone pushing their INTEL k chips...that we largely market/release primarily for this purpose. XD
Sounds like a class action lawsuit just waiting to get started lol.
With the direct statements from Intel employees that clearly and precisely define what "in-spec" is?
Yeah definitely.
Intel's spec is really quite simple:
it improves benchmark scores > in spec
It crashes > out of spec
I’m assuming this now invalidates any previous baseline benchmarks performed, maybe?
If Intel at least using the baseline profile in their presentation then I can see it kinda okay (but really not okay since they did allow it), but they didn't. For example, the Cinebench that Intel posted shows that 14900K multicore is 1.06 faster than 7950X and from what I see, 7950X scored around 2050 to 2090, thus Intel with 1.06x multiplier should score 2173 to 2215 which it can't achieve without using more than 254W! AMD does use PBO in those benchmark, but from looking at The Verge benchmark, 2058 score can be achieved with the CPU using 224W of power (within AMD spec of max 230W) while their 14900K score is 2194 using 324W of power which is definitely above spec and again, this is the number that Intel use in their own presentation! And no, Intel is not comparing actual baseline (as in purely non OC state, thus 125W) vs AMD actual baseline because if they did, 14900K will lose by a mile.
So basically Intel is being hypocritical. Right now if you buy Intel CPU expecting to get the performance shown in Intel's own presentation, you're not getting it with baseline profile. In other words, they are lying about their performance to those costumer that purchased 14900K (or any other K model), running their processor out of spec just to get the win. For Intel to not be lying, they need to allow mainboard partners to default on these unlimited power setting and keep replacing any CPU that run into instability because of it. They can't just say "run the baseline profile if you encounter instability" since with baseline you can't get those numbers from Intel presentation, thus lying.
Push it to the limit
Walk along the razor's edge
But don't look down just keep your head, or you'll be finished
Open up the limit
Past the point of no return
Reached the top, but still you gotta learn how to keep it
A+ I sang it in my head
I have a 13900k and didn't know this was an issue until a few weeks ago. I manually tune my system to stay within my comfort zone for silence and under 80C so that might be why I don't have any issues.
I'm glad this is being brought to light and hopefully makes intel rethink their view of just pushing more power to make it look good.
Hopefully I didn't jinx myself because I said I don't have issues... Would rather not have to deal with a CPU/motherboard swap haha.
Same here. First thing I've done after setting up my 13900k was to limit turbo to 253 Watts manually, I've been running it like that for two months, recently undervolted the core -060 mv as well. Its quiet and trouble-free since then.
@@lenscapes2755 I think I pushed mine hard for a bit seeing it sip 350-400w doing some 6Ghz silliness. But that was pushing the temps up to 90-100C (liquid cooling barely keeping it in check)
For daily I it's at 300w PL2, and 256w PL1 and can't recall my undervolt setting. But with that it typically sits around 80C when fully loaded in my hotter than most peoples rooms. (75-80F)
This chip really needed the undervolt as that alone lowers the temps by 10C-15C for me. Which in turn lowers the power draw.
I'm excited to see how the CPUs of 5 years from now perform. As long as this CPU doesn't crap out on me I don't see me upgrading for at least that long. Maybe even longer if the market gets stagnant like the old days where it was just quad cores for years and year with barely any changes.
I keep the 4096w but use BLCK 100% so clocks only go up to 5.5ghz. Maybe I'm wrong but it seems to me it wont draw nearly as much power this way but is still able to provide enough power to keep stable. I have experimented with 330w limit though but found some instability. I would imagine the locked clock speed keeps it from pushing too hard.
I think intel has known about this issue and just didn't care. Bottom line is no one should have to fiddle with the default BIOS settings for long-term stability. Under-volting for lower temps is fine, but it should not be the go-to for stability in a non-OC situation.
@@darreno1450 💯
It's super ironic that Intel highlighted the "potential for performance degradation" when it was apparent that Zen 4 ran hot
Thanks you very much for all those technical infos!.
Hey, one other angle is with this type of test one finds what games are frequency sensitive so one can refine or go deep in the methodology!
I've always run my 13700k on asus with the multicore enhancements off and a small undervolt, because at defaults it would hit thermal throttle basically instantly under any kind of load. Intel went nuts with the power stuff on these.
BIOS default enables MCE is the biggest problem on all motherboard manufacturers.....
normally home user won't touch bios setting actually......😢
what is mce? really really worried that my 13700k is going to die now 😭
@@boy-who-likes-bats Multi-Core Enhancements, It's what Asus calls their default over the top power settings in the BIOS, other motherboard manufacturers have the same thing but presumably by other names. Switching off MCE sets everything to "Intel Stock" power settings but.... as mentioned in all the coverage, what actually is a stock setting is pretty vague. I never dug into exactly what turning off MCS changes, but at the very least it sets PL1 and PL2 to 253w, similar to the "Intel baseline profile" I can't recall seeing what it did to other things, like the svid behavior, iccmax or whatever else. But at the end of the day, all the reports have been around the i9s. I haven't seen any reports of issues with the i7s, and I'd imagine many more of those have been sold, so problems would have higher visibility.
@barrym426 ah, hm. i don't remember if i touched mce when i set my pc up for the first time, but i do remember pl1 and pl2 being 4096 at one point, but i've never seen cpu wattage go over 250 during any load, synthetic or gaming or otherwise. it's been stable and reliable for me for now, i know my chip runs a bit slower than a lot of the oc guys have theirs, but it doesn't get much hotter than 80C, rarely ever.
just really not looking forwards to having to buy an entirely new cpu if mine decided to die one day.
Who to blame: 70% Intel, 20% Board Manufacturers, 10% Reviewers for not doing power-normalized benchmarks by default (which motivates Intel+Co to do this stuff)
What are power-normalised benchmarks? Top Gear doesn't limit supercars to 500hp for their lap times, why should reviewers start limiting stuff?
Can't blame reviewers and that isn't fair if they were included. They do their work with the impression that "stock" means out of the box, pop in and run tests. As a consumer, I appreciate that more than for them to go in and make any adjustments. People who buy these parts are not rushing home to tweak BIOS settings to "power-normalize" anything. Imagine the outcry if reviewers were to make changes prior to their benchmark findings. Then they would be labeled shills for their preferred pc parts to perform better.
@@mazing87 Of course they should still also do benchmarks with stock settings. The issue is that most reviewers only do those. Due to that issues like this linger in the background, and only get real attention when things blow up.
The truth is (as can be seen here) that at a cost of 5-10% peak performance Intel could reduce their power consumption and heat dissipation by 20+% (and any user could easily do that by customizing the limits, but very few peole realize that).
@@renerant Yeah, why the f does GN do noise-normalized tests, they should just ramp the fans up to 100% ...
The fact that you assigned any blame to reviewers means you're wrong.
fantastic coverage, thank you. In particulat 8:36 highlighting how much performance AND importantly, power usage changes with the more stable profile.
Ohhh spicy video, can't wait for certain comments to bash these findings
To bash the findings you'd have to ignore the facts, so you're right it will happen :D
Yeah. There was a comment made immediately after the vid was uploaded ...a 25 min video. They couldn't have watched it.
@@domm6812Intel fanboys are the most pathetic. Amd fanboys in my experience are a bit less annoying overall.
you will see those comments in X lmaoo
Wonderful reporting and clear explanation of this issue. Thanks for your hard work as always!
Isn't this the same craziness that you were banging on about when you were monitor Steve?!
Modern systems man.
Go with Intel and you're not running hardware in spec, go with AM5 and your system boots up like a 1980s Compaq or go with NVIDIA and melt your mother flipping 12VHPWR connectors off.
Am5 has solved the boot up issue went from 2 mins to like 20 seconds for me after the updates
Great video. At work we've been implementing our own version of the baseline limit for almost 2 years. Even the best waterblocks couldn't keep up with these chips when unleashed. Its been frustrating having to set customers expectation on 10-15k water-cooled system they can still see 100c in same situations. Some convert to 7800X3Ds and thank us after the fact. but those who don't some always complain even after we set their expectations. We're a little scare for what's to come with the next gen flagship intel cpu/nvidia gpu combo.
Whats the differance between the new "baseline" profile and the old "enforce all limits" profile?
I remember people (including Intel) gave shit to AMD for the 9590 drawing 300w on a 220w TDP. Here’s intel drawing 400w on a 150w tdp on a chip that lives its entire life above 90c stock when playing a cpu intensive game or high refresh rate setups.
You can’t even cool a 14900ks to maintain max clock with just a custom loop.
Assuming electromigration is causing the degradation, current is actually the issue, not wattage. A fast corner CPU could run at a lower voltage but higher current and be over the damage limit while under the wattage. Eletromigration doesn’t really happen until you cross the current limit, then it’s exponential.
This was a clear demonstration of the issue, and it cleared up a lot of questions I previously had. Awesome work
This is intels fault. You can’t say Intel is innocent because the board makers are boosting beyond defaults - because Intel isn’t enforcing it. They want the higher benchmarks, but blame the mb for the stability. Trying to have their cake and eat it too.
It’s same as Nvidia and the power cable. Try to blame the consumer when your faulty designed cable allows for user error.
Edit: haha. I wrote this before watching the video and Steve says the exact same thing at 17:22.
Thanks for this video explaining the situation.
It's interesting to see again just how far the power curve these CPUs are being pushed. It was cool to see the 14900K running at less than half power and half the temperature and getting 75% of the performance. I loved that AMD acknowledged it from the get go with Eco Mode for Ryzen 7000.
I initially resolved my issue by downclocking with P-core ratios, but this solution didn't feel ideal. After downloading the latest MSI BIOS with ME 125, I still needed to adjust the P-core ratios. Modifying PL1, PL2, and other settings didn't seem to help.
Recalling Intel's statement about a voltage issue, I decided to set the CPU LLC to Intel default, reset the P-cores to auto, and restored all other settings to their previous values. This approach worked successfully. I tested it by running Edge RUclips, Claude, Intel Process Tester, Cinebench, and other applications without any problems.
It's worth noting that with these settings, Cinebench R23 scores decreased by about 3500 points compared to when I had set the P-cores to 54 on all but two cores, with those two at 56.
However, I would rather have less performance until the voltage problem is addressed. This approach prioritizes system stability and longevity over maximizing benchmark scores.
I really wish at least one motherboard partner would have came out and said we cannot in good faith support the unlocked 13th and 14th gen CPUs. There is no way to build a motherboard and a cooling system without going to excessive extremes to keep these processors safe.
Currently have an Intel i5 14600k in my current build. So far so good.. (Fingers crossed)
"We're going to be very crisp in our definition of what the difference between in-spec and out-of-spec is." You know what else is going to be very crisp? The CPU, after running at those wattages.
This happened to me the first time I got an intel i9-14900k! It was working perfectly for a while. Then I applied what I thought was a mild overclock to it (nothing crazy, just enabled the AI overclock feature) Than it all went downhill from there. The next day it was blue-screening after launching a game. So I turned off the OC, but it was still blue-screening! I wound up replacing almost every component starting with the RAM first before I finally caved in and returned both the CPU and MLB... I had no idea that these processors are so unstable, just recently I watched Jayz-two-Cents video about the same topic and disabled the default BIOS settings to "enforce all limits"... I only saw about a 3-5% performance hit, but my temps dropped nearly 20 C!
"when you sacrifice stability for performance, you deserve neither"
- benjamin franklin
thank you for clarifications . 🙏 👍
One of the takeaways is to continue to avoid Gigabyte.
My 13600K is on a B760M Steel Legend and I have no issues. ❤
Can we get stability tests between the 3 boards/profiles?
With my chip it's pointless, my chip is stable using unlimited power, probably because I've not used it much (or it's a good chip).
@@Hardwareunboxed similar results with 2 of mine. But I've been running mine at or near 400w for a year with no instability. Running an EVGA KP z690 and Asus strix z690. In fact the only instability I had was going over 7200 xmp on ram. Strix is running 6000xmp and EVGA is running 7200 now and it's solid. I wish we had actual data from these unstable setups because I think there's more to it. Things like ppl running 4 dimms or "Hero" voltage to try and hit 8000m/t on 2 dimms. Sadly the pitchforks are out and the villagers smell blood so we'll likely never get real results.
This would be the real test. With how much outlets love to clickbait this shit you don't know if it's 50 CPUs or 50K. None of the i7s or i9s me or my brothers have built have had stability issues, and I don't personally know anyone that burned up a 7800X3d or 4090 either. Doom and gloom gets clicks and Intel is a juicy target to take shots at.
i recently did this, Asus Z790 was worst, cinebench didn’t even start
@@zodwraith5745exactly. How wide spread is this issue? Releasing the bios isn’t because everyone is failing. So again it’s unknown I guess.
Will you be doing a fresh set of benchmarks of the 14900k vs the 7800x3d now?
Why 7800? I want to see 7950X3D.
@@paradoxicalcat7173 either works.
Most people just consider the 7800x3d the best amd cpu for gaming since it doesn't have issues with having to ccd.
I wonder how a certain chaser youtuber will react to this situation, unlike a mythical nonexistent problem with amd he made up, this thing is truly can be called Inteldip
Too busy scamming his audience 600 bucks for unstable intel memory overclocks.
@@blegi1245nah, he’s just too ethical to do any such thing. His principles are too high. He’d never lower himself to that sort of behavior. He says this about himself in his videos. We should all believe him when he says things about himself like that that no one asked to hear. 😂
What did i miss 😮
I have a 13900k that I ran perfectly stable (with extensive testing) for over a year. Then suddenly it started crashing a few times a day. Then it became every hour. It wasn't an extreme overclock (in fact, I was undervolting slightly, and running with power limits). I'm pretty sure it's the same issue as whatever is being reported here, because the only way I could resolve it is lowering the clocks. The strange thing in my case is I *did* constrain the power limits to within spec.
Not much changed with my 13600K 55x43x48, 4000cl15 ram. Only had to up the voltage by 0.010mv one day after some windows updates.
Same here with a 13500 on the MSI board, crashing and freezing. The 13700k with an Asus is still fine but I'm worried about it now
In the video he did question whether "in spec" 253W is still killing the processors. My gut says YES, because physics. Until recently, all CPUs were limited to 90-120W TDP. There was no sudden overnight break-through in materials or cooling, so leaping to 200+ W screamed to me that there was a problem.
@@pixels_per_inch Freezing is unstable memory controller, crashing is too much vcore drop.
@@paradoxicalcat7173 I locked my 13600K at 160 watt. You have to be mental to go 250 without delid and custom loop. Now I'm pushing 5.6 P-core 4.4 E-core 4.9 Cache and still fine.
Wow, and some Intel 13900k fans telling me choosing the 7950x3d for lower power consumption doesn't matter... Lower temp, no throttling almost, stable (and now just for 1 year), lower power bill, less expensive cooler required, might last 7 years unlike Intel i9.
I'm sad for Intel customers.
I've been watching the 12/13/14th gen users complain for ages about stability, and how 80+ deg. C is "normal" and "OK". AFAIK the laws of physics never changed, and neither did the materials used in the processors, so 80+ deg. C is getting very close to melting stuff. It is now making sense why these things are running so hot and so power-hungry. It's not that they need to be; it's because Intel want performance at literally any price. The price in this case is system stability and greatly reduced component life.
At this point, I really think that going for non-K SKUs seems to be a better bet. I mean, have a little less clock speed, quite a lot less power usage and still good performance.
Intel CPU's are fine all motherboards are set at 4095 watts which needs to be changed nothing to do with Intel ...This has to do with who is building the PC is at fault not intel and i9 13 or 14th gens carries a wattage of 253 watts on the PL1 and PL2 and it should be set at 307 amps for that CPU also for other intel CPU's these numbers would be different u can get the specs of the CPU on intels site unfortunately a lot of people building these systems don't like to read and set the CPU at spec on the boards so don't blame intel people that build the PC's should be blaming themselves cause I don't have any issues with there CPU's cause I set the numbers as per spec.
ruclips.net/video/OdF5erDRO-c/видео.htmlsi=p6ISU1bqKV0bTZHm&t=1123
@@Hardwareunboxed there is no issues I've been running my CPU since it was launched never ever crashed cause I already put the right specs in the board I never go by bios defaults that's not how I build PC's all boards are defaulted to 4095 watts not just 13th or 14th gens all of them as I always check no problems here my friend everyone is making a big thing out of something that people should be checking
Saying your PC is fine therefore it's not an issue is extremely tone deaf. This issue is likely only affecting the poorest quality silicon, higher quality silicon will last much longer, if degradation is the issue.
The values you mentioned for PL1 and PL2 that the gigabyte motherboard is using are the recommended for the i5 14600k.
And some people still have the audacity of saying that Intel is more " stable " and " reliable " than ryzen.
Or the old "AMD CPUs are run hotter than Intel...
I'd say, yes, if you leave it the hell alone, a Raptor Lake CPU on Z690 is staggeringly more stable that a Zen 3 platform. My 13600k has literally never BSOD'd or otherwise hard-crashed. Sure, the occasional game will inexplicably crash ( _maybe_ once in three months), but that's just bad coding by the devs.
@@awebuser5914 It seems like the i5s are not really a problem and the 13600k or 14600k is probably the best all around CPU for the socket for the money IMO. I have a z690 and z790 board and will happily run the i5s until I need something else. The i7s and i9s are a problem though and Intel needs to fix it right or they better have a good showing with 15th gen.
I never had any issues with intel CPUs, and i dont have with current 13900k, i dont even know how people manage to push this cpu to its limit in a unreal engine game
@@awebuser5914 anecdotal at best, assuming you have both systems and know what you're talking about.
Very interesting discussion. Well done.
I already knew how to adjust the power profiles in the BIOS, but just for giggles I updated the BIOS on my z790 Dark Hero. Originally current limits were flat out unlimited. Intel published new profiles showing 307 amps and 400 amps. When I updated the BIOS, the Intel default current limit was set at 280 amps. This whole mess is all over the map.
I'm so confused about the 280 as well.. At this point, I don't know how to adjust the settings to make it stable.
@@devans83 I haven't had any issues at 253W with the 400amp Icc max. Any kind of overclocking and I have to walk a very fine line. That said, I don't have any problems running the E-cores at 45X just for a little uplift on the cheap. I'm also running a slight undervolt on the cores and the cache domain. usually -0.15 to -0.20. Your mileage may vary.
Thank you for diving deep into such issues, showing what is actually causing this whole issue. Seems like the only way to fix this problem is reengineering the CPU core architecture in order to achieve high performance at a relatively lower power level...
I like how soyo and maxsun have better temps than anything else
I am using a Soyo (basically a rebranded Maxsun) AMD B550 board. Other than the initial pain with the BIOS (am not sure who's at fault here, me or Soyo), it's been running excellently with great temps and stability. Decided to try out Soyo because of good memories of the original Soyo back in the 90s and also good reviews from youtubers about the board.
@@DonaldLim-r2t I never heard of either of them until now, from what I can tell they both must be fully chinese brands because they seem to only be available from aliexpress, but both are worth to look out for.
This happened to me as well. I built a new PC in January and my first combination was an MSI Z790 Tomahawk paired with an Intel 14900k CPU. After about a week or so, started getting random crashes which got worse to the point where i would get the out of memory error when booting up games, such as Hogwarts. I did a lot of troubleshooting, which I will spare the details here, and ended up switching motherboards to the Asus Strix Z790 Gaming. This was not the issue. I ended up replacing the CPU to another 14900K and this worked immediately, however, the problem came back after 1 week of use... I was lucky that I was within my stores return policy. That said, I couldn't trust Intel any more and swapped everything out and built a new system around an AMD 7800X3D chip, which has been stable ever since, albeit not as fast. I thought it was a bad batch of CPU's, but now seeing these reports, it sounds like this issue runs deeper.
Seriously, what is the point of buying a premium motherboard if you can't take advantage of it pushing the CPU???
You completely and utterly mis-understand the problem. The problem is NOT the motherboard. The problem is Intel running their CPUs way beyond what the silicon can handle.
@@paradoxicalcat7173 I'm pretty sure I was clear in my comments that it was not a motherboard issue. I was simply stating that during my troubleshooting, I ended up swapping motherboards since I couldn't figure out the source of the "out of video memory" problem. Having built many systems in the past, this was my first time experiencing this type of degradation in a CPU and didn't think it was possible without a hard overclock. I ended up trying 2 x brand new 14900K CPUs and both of them failed after I installed them in 2 higher end motherboards, running at their default settings with only XMP enabled. Intel has confirmed in the past that this is not "out of spec" to have this enabled. My last comment regarding the point of buying a higher end motherboard as being useless, is because these motherboards push the CPUs to the limit, which obviously causes some to degrade prematurely... If Intel now points to these motherboard manufaturers as being the problem, and need them to remove any type of "boost", then you might as well buy a basic motherboard. That's all I was trying to say...
This only makes the KS situation a lot worst, wow.
And they have a bunch of them. 13900KS, 14900K, 14900KS, are just bins on top of the top bin. For this, you get the privilege of supporting the company with your money.
@@randomguydoes2901 LOL
Remember when auto-OC became standard and people all over the internet were concerned and questioned if this was safe for their CPUs and their concerns were dismissed because of all the built-in safety features and even manual OC was dismissed as unsafe compared to auto-OC?
Thanks man, would be nice to see the gaming benchmarks of the 7800x3d vs 13900k/14900k with the new baseline bios updates.😀
7950X3D. 7800 is meh.
well i havent had any issues with my 14600k on my asrock z690m itx but that board also limits power to 125W sustained so maybe thats why
Oooh cheeky wee sunday morning one. Its like when GN listened to ASUS and sarcastically ran the AM5 socket completely stock everything and showed how performance was lowered 🤷
MSI boards have had the option to use Intels reference power settings. BIO menu go to advance - > OC - > cpu tuner setting and select box cooler.
After 30 years using Intel I got my self a 7800x3d and i am a happy man.
thanks for the vid, it would have been great if you included the 7800X3D and 7950X benches to the table for an easier comparison, may be in next vid.
meanwhile in MSI, they just said to set CPU Cooler Tuning to air cooled to limit it to 253w T^T
It means that they are confident with their BIOS settings.
As someone with a 13900k and MSi board I can say it works. You get similar performance to the ASUS profile shown here for 253W and it dials back the voltage and heat quite a lot. If it's at 4096 it will push voltage above 1.5V and will hit 100C. At 253W it's a 10-15% drop in performance, hovers around 1.38V and it runs at like 80-85C. I think MSI bios acts different to ASUS and Gigabyte but I could be wrong.
They clearly did mentioned another step to select “intel default” under CPU lite load control option. But this will use higher voltage compared to MSI’s default option. Not sure whether it’s better to switch to intel default for this step..
Thankfully I've been running that since day one and it runs like the Asus boards profile in this video
17:50 Given the power use and process node it seemed possible to me Intel was trying to max their chips out to compete with AMD at 5nm. Seems very likely now. Not cool to given high spec CPU users are now paying for it.
How is motherboard BIOS settings that push a CPU past its intended operating limits ,which results in the CPU crashing, Intel's fault? My 14900KS is perfectly stable, and has been from day one... BUT, it depends entirely on what revision of BIOS your using. Im running at 8000mhz 24hour stable memory speeds and the CPU is functioning with zero stability issues.
It's Intels fault mainly because the intel baseline is not usually the default setting out of the box. Intel made the CPU so they should be enforcing the baseline power limit as default. Any changes to this by the user is then the fault of user if the CPU degrades. AMD just had a similar issue with 7000 series CPUs frying because they didn't make board partners enforce a voltage limit and let the board partners push things way too far.
Intel and AMD really must watch these board partners from now on to make sure they don't push things way out of spec because it's bitten both of them in the ass.
The motherboard manufacturers have to sign a contract with Intel to get chipsets and firmware to build motherboards. So Intel has all control in the world over this. But they want to write "TDP: 253W" on the box of the CPU (so it doesn't look too bad) and then have the board blast the CPU (with plausible deniability) to win benchmarks.
So yes, it's Intels fault. Very simple. Nvidia doesn't allow their partners to just do whatever with their GPUs (much to the frustration of EVGA, but that's a story for another day). To get an OC VBIOS signed by Nvidia (otherwise the OS driver won't talk to it) the board partner has to send Nvidia the VRAM layout and component choice along with a standardized performance test of the cooler. Intel has the power to do this too. But they chose to make it very opaque what who is actually defining what "TDP: 253W" means in any particular case.
For reference: Not a single professional desktop machine from HP, Lenovo or Dell runs Intel processors differently from Intels "guidance values" (And trust me, I've checked a fair few). Because corporate decision isn't made on benchmarks in popular media. In this environment power consumption actually matter, and 5-10% performance difference gets easily washed out in service contract details. So here Intel are magically honest and proper professional. Who would have thunked!
Still thinking about that floating mug at 6:12
(yes its probably something behind processor hidden that its on top of but don't ruin the fun)
One of the big problems of relying on online reports of crashing games it that games often have serious bugs anyway and it's hard to distinguish between the game crashing because of it's own problems or the underlying hardware.
This is true, however there's a lot more to go on here with multiple companies conducting investigations and all pointed at K-SKU 13th and 14th gen CPUs being the problem.
@@Hardwareunboxedmy AVX 512 enabled i7 12700 non k with asus tuf z690 plus wifi d4 at the cost of single i9 12900k in india is still going well since launch jan 2022 😊
@@CaptainScorpio24 Glad to hear it -- but... the issue seems to be with 13th and 14th gen CPU's though... not 12th.
@@CaptainScorpio24You dont use the intel Ddr5 controller
@@predabot__6778And it's K-series CPUs that suffer from this, since those are overclockable and the motherboard basically overclocks them out of the box.
My i9 13900KF, which is now over a year old, was a champion at the beginning, without any issues. However, now, without enabling the Intel Baseline profile on my motherboard, I crash quite frequently whenever I play Remnant 2. I faced this same issue previously as well, but it's worse now. The Baseline profile fixed this, but at the cost of performance.
*Take this with a grain of salt. I overclocked my CPU a few times, using my motherboard profiles and a bit myself with the help of professional RUclips videos. Though in my humble opinion, it didn't get worse, but in benchmarks, it would be unstable, even without the OC profile. However, a simple reset of my PC fixed that.*
If your pc continues to crash in benchmarks you need to adjust your motherboard's LLC calibration because your CPU voltage drops too much under higher load. find out which profile is stable and doesn't overheat your CPU
HOLY CRAP! Antec! i havent seen them for ages. Good quality.
let's see what their shot at an o11 clone can do
Great investigative journalist @HardwareUnboxed thank you for such thorough coverage!
I am so glad i ordered amd 7600x to upgrade from i9 9900. 🤗
I think the 9900 was far from obsolete for it to be upgraded should've went for the 7800x
For what it's worth, these problems only happen to the top end intel CPUs
@@MiGujack3 7600x is still much better for gaming then i9 9900(nok K) and i just wanted to switch to am5 with affordable option atm. Next upgrade will be zen5 9000 series AND i don't have to change my board.
No real brainer for me for sure.
@@samarkand1585I have a 14700k and it’s a problem
I ❤ my 7950X3D. Wouldn't go Blue again, period 🔥
I have this problem for almost 3 years ago for intel CPUs in my workshop from 12th gen to 14th gen because it consumes a lot of power and my customers complain about the heat it goes up to 95 degrees in any 360mm aio or 420mm liquid cooling system 😢 and i decide to lower the power consumption of the CPU or limit the overheat temperature protection to 80 degrees and that's works for me some customers doesn't care about the frequency of the cpu they care about the heat and some other care's ...I tell them the performance will be less about 7 % and they are satisfied for that ....everyone is happy ...they happy 😁 me happy ....for every build i build it for intel CPUs and I believe the common enemy for electronics it's the heat
More power = more heat
More issues..unstable.. crashes ...for me it's like overclocking unless you have the best power stages or best VRM components and best cooling system to gain this much power
Intel cannot command motherboard makers do anything but in the Intel white papers you can see the power limits. Page 13:46 in video. This is why both Intel and AMD CPUs should be benchmarked at stock, with stock RAM. Ask yourself which companies dont have to change their benchmarks? The ones that do have to change their benchmarks because of this are the ones that dont know how to benchmark correctly. AMD CPUs are massively overclocked in benchmarks as well. AMD CPUs were burning IO dies to death because the motherboards were putting way too much voltage into them. Yet benchmarking sites cant read the documentation and do stock benchmarks. So we have extreme overclocking on both Intel and AMD as standard. DDR5-6000 RAM etc to make AMD cpus faster, so you get the maximum out of their IF @ 2000 MHz. More or less the maximum you get out of AMD. This is massive overclocking as well. Reviews are the problem, their benchmarks are all massive overclocks. Also Intel they dont increase the cache frequency like they do with AMD's IF. AMD's IF is maxed out. Both together on both CPUs equal a big bump in performance.
I blame sites/channels like Hardware Unboxed. Its their biased benchmarks that lead to motherboard manufactures maxing out everything. Stop overclocking your Cpus in your benchmarks and watch the problem go away. Stop blaming Intel because you dont follow their documentation. Look at the state of the AMD overclocks in your reviews. How is that fair, their IO dies failed due to the voltage of getting those IF@2000 and DDR5-6000 overclocks which you used in all your reviews. Also stop pretending to be experts, your not. If you were you wouldnt be rerunning to see the performance loss. Dont act like its the fault of someone else as experts you would have limited the power draw of Intel CPUs and not overclocked AMD CPUs to death. Increasing IF to 2000 and using DDR5-6000. Really biased for AMD stuff. Hopefully AMD payed you for your reviews with free hardware. Maybe they told you how to setup the system.
If you can find a benchmarking website that didnt have to change their benchmarks. Then you know they are unbiased. There is an overclocking channel laughing at all this, because they could read Intel documents and saw this comming. Same with AMD's IO chip getting way too much voltage. For us overclockers, we know you cant trust benchmarks. You know results close to your maximum overclock on a silcon lottery winner cpu. AMD CPUs are by far the worst for special tuning before benchmarks.
From Intels website Intel® Core™ i9 processor 14900K :
Processor Base Power
125 W
Maximum Turbo Power
253 W
Memory Types
Up to DDR5 5600 MT/s
Up to DDR4 3200 MT/s
Were your benchmarks running these power limits and memory speeds?
From amd 7800x3d
Default TDP
120W
Max Memory Speed
2x1R
DDR5-5200
2x2R
DDR5-5200
4x1R
DDR5-3600
4x2R
DDR5-3600
Were your benchmarks running these power limits and memory speeds?
If not the problem is your benchmarking.
All the power specs are in the datasheet, volume 1 of 2, including Extreme Config. At no point in the datasheets does it say “no limits.” Unless Intel writes it down somewhere for the public to see, they’re covered.
The “baseline profile” varies from CPU to CPU. It is not a one size fits all setting. Gigabyte is using the 6p+8e 125w CPU settings instead of setting the values for an 8p+16e 125w part.
This isn’t rocket science, and MSI has handled it for years. On a new CPU install, MSI asks what cooling is being used - box, air, or water. The board then sets the power limits accordingly.
I had a MSI build and it defaulted to "water cooling" with all the limits removed. In fact none of them are Intel specs, both tower and water cooling are both unlimited at 4096 Watts with only box having a sensible PL2, but PL1 is set too low at 65 Watts even for unlocked CPUs
Most of my friends on Intel are running the i5 12400 and haven't upgraded. No E-cores so no need to use the application optimization software, and no crazy power consumption.
This is definitely Intel's fault. They were just about encouraging motherboard makers to push these CPUs because "bigger bar better"
188W looks like halfway between 125 and 253 (minus 1 watt) and thus like a compromise. There is no other SKU that has a PL2 as default, so not likely an error recognizing the chip or something like that.
There used to be a formula for PL2, like PL2 = 1.25 x PL1. That was too constrained for the higher core count CPUs so now it adds a 'nice number' (128) to the 125W 🙂
I understand Gigabyte for playing it safe here...
i feel happy with my beloved 9900K
I was looking for one of those recently. In New Zealand people pay about $400 NZD used, while the 12400F is $240 NZD new, and it out preforms, with much lower temps.
Upgrade to 4080, 4090 and you will no longer be happy.
Wait so what about those of us who enjoy gaming but aren't as "in the know" in regards to this stuff? I've got an i9 14900k and an asus motherboard, while I haven't had any issues is there anything manual I need to do in order to future proof?
Yes just disable ASUS MCE