@@sebastianandres6802 No there is a permanent fix. The bulk of their CPUs can't actually handle having boost clocks at 5.8-6.0Ghz, because the voltage required is too high to not cause degradation. Max boost should probably be 5.5-5.7, which is too equivalent to AMD to have a competitive advantage.
I think their biggest lie is stating flat out that the previous microcode was somehow acting in error, or giving "incorrect" requests - it was doing exactly what they wanted it to. Their motivations were purely to beat the 7950X in benchmarks and advertise a boost clock of 6Ghz, and they tweaked everything they could specifically to squeeze every last drop of blood from a stone essentially.
If I got that right, then the standby voltage was too high as well. This is not in line with your assumption that they did it on purpose. Still if they did it on purpose like you said, they made an unintentional mistake by giving to much standby voltage.
@@zCaptainz I just stated what I heard in a different video, not sure if it was Tech Yes City, Hardware Unboxed, Gamer's Nexus - probably it was der8auer. I don't own any Intel product and don't care about their economic performance.
If Intel was a body shop, they would paint over rust. I'm an ICT architect and my brain doesn't fathom Intel silicon being non-compliant with manufacturer specification when they are the manufacturer
You misses the part where Tech YES City got it all wrong. Microcode works correctly on undamaged chips only. Pulling out a CPU that was acting up for testing was pointless.
@@Reeper-x6x Yeah lets fix a broken CPU, that's cooked.. it runs, but its core phreaked, over peaked n wouldn't even be worth throwing on eBay for the grief...
I was having my lunch at the time, wasn't even looking at the screen and almost spilled everything out, could not stop laughing. I had to rewind to listen to it again. Let the class action lawsuits begin😂
@@gingerbread6967 The possibility for a class action lawsuit is already being investigated. Intel has known the issue for at least 2 years and it is not difficult to prove it in court. Tech Yes City is not ''misinformed'' alone, Level 1 Tech, Actually Hardcore Overclocking, Gamers Nexus, Hardware Unboxed, Jayz 2 Cents, and a few others are also on the ''misinformation'' train. Lots of ''misinformation'' going around these days.
They even doubled down on these issues by releasing the second flawed generation. I dont think someone with rational thoughts is making decisions at intel. Since sooner or later it will cut into their own flesh. And the damage to the brand is indeed massive. No one in their right mind would ever by Intel again after going through this mess.
I remember when it was the end user who used to overclock, now we got the main manufacturer overclocking (to the point of degradtion) and it is the end user who is now dialing it back... strange world we live in.
What's crazy is that they don't even do a good job of it. My Ryzen 7600 out of the box with a big air cooler and PBO enabled used to run at 85c and couldn't hold full boost frequency on all cores at the same time. Then, I let the Ryzen Master curve optimizer do it's thing. Now it runs at 65C and holds full boost on all cores at the same time. It gained like 300-400 CB points too. Either I won the silicon lottery or PBO just does a horrible job with stock settings.
@@madmatt2024 For what it is, PBO is incredibly useful for how little effort it takes, even if there could be better settings. it's not supposed to automatically fine tune for your CPU, but it allows for you to experiment and do that.
PBO is the worst implementation cause they want it to age the cpu more faster so instead of lasting a long while it dies prematurely. On account of that every1 should disable pbo. I found an article that benchmarked with pbo off and it really doesnt gain the cpu almost any benefit to have pbo while the cost is like 20c extra temps and downclocking during load
I mean, not really. Of all the reasons I don't want chips in my body power draw isn't even a factor. Your brain runs on about 15w, I'd imagine there's not much energy to spare lol
13700K, disabled e-cores early on, no issues except a random crash here and there playing Apex (but that could also just be an Apex thing), and that latency on opening certain folders and video editing that I've been complaining about across a couple of Intel CPUs now (13700K and 10980XE). The PC is almost 2 years old now, so... should I be expecting issues later or is it possibly ok?
Yeah the 10980XE is actually worse than the 13700K when it comes to latency, don't know why. Once Tech Yes is back in Aus, maybe he can do some tests and figure it out, could make an interesting vid. And yeah the 10850 and 11900 are my 2 most stable, best performing PCs at the moment. Had I known this was going to be the case, I wouldn't have bought the 13700K thinking I was "upgrading". Going to test the 7800X3D next to see what the overall performance is like compared to the others.
@@RocketJumpNinjaI purchased a 13700k and I RMAD my gpu 3 times thinking it was my gpu , at one point I’m sure GPU was an issue but now I’m starting to think holy crap it was my 13700k all along wow. I think I’m rebuilding my 5900x to test stability with my GPU.
Thanks! Please let me know if my interpretation is correct: The new Micro codes Increase voltages & increase the allowed thermal temps. Those two factors eventually throttle the CPU because of unsafe voltage and and/or unsafe temps. So in the end, the new micro codes are enabling the CPU to reach the originally advertised speeds but degrading the CPU at an exponential rate? right?
Yes if the cpus are degrading then this seems like a way to have them still work temporarily. It's minor in difference but I do expect potentially another microcode update in a few months. I mean there has been actually heaps of them for the 13900k. Some concerning stuff.
@@techyescity I had my PC built by Micro Center because they stand by their customers. I told them how the new Micro Code is the equivelent of Gai Sensei opening the eight inner gates except nowhere near as epic. My 14 series based PC was thermal throttling on their test bench so hopefully they will make it right in some way.
I get it. Intel wants to avoid a class action lawsuit of reduced performance on these parts. They increased the voltages to avoid the crashing issues at the expense of chip longevity since they expect not too many people run these 24/7. If a CPU dies under warranty, they would happily replace it until it runs out of warranty
The excess voltage happens on degraded chips, Sadly, a Tech YES City used a CPU he knew was faulty to test. You can't fix a bad CPU, period .It need to be RMA'd
no, u guys understood it wrong. the microcode fixed the voltage spikes that went up to or ABOVE 1.6v & thats what killed the cpus. that happened really quickly u usually dont see unless u have a graph that monitors every tiny voltage change / every ms & thats what the issue was. Intel cant fix already damaged / degraded chips. thats why they offered a 5 year warranty for every chip that was affected by it. (if ur unsure if its damaged u still can make a warranty claim to get a brand new one, or in other words just lie & say its damaged.)
Yes this is probably the legal game. They have marketed those speeds but not voltages and durability. They will not give a microcode update which includes lower speeds since it opens them to litigation. They'll just juice up the CPU at different intervals to prolong the lifetime until they're out of warranty... gg thanks for playing
If no one complained, Intel will never take action of this. This is why we should be vocal against billion dollar corporations. Thank you TECH YES CITY and other tech UA-camrs for your efforts in the tech community.
I think it's crazy that you pointed this out a year ago and that largely went unnoticed, but I remember watching those. ..Bravo to Mr Tech Yes City. Hats off to you sir
I have 11th Gen I5 and it works fine. I grew up on AMD Intel was feeling like snobby CPU at that time I grew up on. AMD is kicking both Nvidia and Intel butts in the stock market.
This issue is not about voltages UNDER LOAD. Under load, the voltage does not go higher than 1.5 volts. It can't, the chip will power limit to under XXX watts (depending on the chip model). This is not about wattage (power draw). This issue is not about temperature. All people testing under load or looking at temperature to see if the problem is fixed and all the people running motherboard sensor data to determine if the issue is fixed or not are not looking at the cause or the fix. Firstly, my understanding is that motherboard sensors are not sensitive enough to consistently measure the actual voltage spikes involved. The sensors are sensitive in the milli-second ranges and the spikes in question are in the micro-second range .. so 1000x faster than the senor is designed for. So, the peak spikes shown are completely irrelevant with the actual spike felt by the chip / ring bus. it is like measuring your blood sugar 5 hours after eating a candy bar. You see and say, my blood sugar is normal. Sure, but it wasn't at 15 minutes or 45 minutes or 120 minutes. But it was great at time 0 and time 300, the only measurements you have. The motherboard sensors take a reading at time 0 then again at time 1000, then at time 2000 (in a micro-second event). They don't see times 1 to 999, times 1001 to 1999, etc. So, if you get lucky and if the spike actually happened at time 1000 or time 2000, then you see it .. if it happened at a time where you did not measure (a 999/1000, or 99.9%, chance if we are talking a micro-second to milli-second event comparison) then you just don't see that voltage. Should you just take intel's word for it? Absolutely not! But wait until some people with proper external equipment and knowledge actually use equipment sensitive enough to detect and diagnose the problem actually do it.
Running CPUs at high voltages for a short time does not degrade them a lot, that's why extreme overclocking works at much higher voltages with the CPUs still working fine for years. What degrades a CPU is a combination of continuously high voltage, high temperature and high current. But while current and temperature increase degradation linearly, voltage increases it exponentially. 1.5V is not a safe voltage to run a CPU for a long time. Not even Bulldozer, which was based on a bulletproof 32nm silicon on insulator manufacturing process, could handle more than 1.45V continuously. On current manufacturing processes with much smaller structures, you're better off not going far beyond 1.4V under load or beyond 1.35V continuous if you want the CPU to last beyond the warranty period.
@@razorblade7108 that is true in general. But THIS specific problem is damage not caused by long term high exposer but short term very high spikes. This kind of damage did not exist before now. It is all about how they wanted quick boost ability. I have no issue running at lower voltages in general. I absolutely undervolt. But that is not this issue and should not be confused for this issue. You can limit to 1.4 and you will still get these spikes if you all boosting.
ive seen tests that recorded voltage spikes up to 1.6v & above under load or in the beginning of the load. so its def. a problem. & after the update it didnt spiked to the 1.6v anymore (also core speed lowered by 20mhz or so). the spikes are still relevant since cpu are voltage "sensitive". i agree with the temp & watts but that should be common sense for the most part - still relevant but not in this case. However not every CPU is a gold sample. every cpu has some "defects", some will run fine at the 1.6v or 1.5v spikes but others cant take those. some clock a bit higher at lower voltage while the next cpu needs 0.05v more for the same speed or it doesnt clock as high as others resulting in 20mhz less or more depending on the chip. + Some motherboards tend to give more voltage than others on top of it. CPU have a margin of errors they can have when they got produced. too many errors/defects disabling certain feature sets/cores = i5 or i3 while the good ones are i7 & i9. some will even work for maybe 5-10 years without the microcode fix without any problem since they prob got the golden sample, others have chips that barely got the i7 mark in production & if u got a chip like that, then gl surviving the spikes long term. Extreme overclocker using usually a golden sample for those high clock speeds & voltage. Motherboards sensors arent 1000% accurate but its the best u currently have for the normal customer and for the most part they are accurate enough to notice that there is some issue or change.
@@Vss077 If those spikes are really high and happening often and long enough, they can add to the problem, but in general running any CPU at 1.5V for longer periods of time will degrade it no matter what, even if it's a golden sample. Also a CPU which is a golden sample means it runs really efficient, so it needs low voltage for the same clockspeed, but actually it could also degrade faster at higher voltages. It is less likely to degrade when running at the same clockspeed with lower voltages though. Regarding the sensors I agree that motherboard sensors are usually not accurate enough, but nowadays you also have sensors all across the CPU, measuring temperatures and voltages more accurately and also enabling them to account for voltage droop between VRM and CPU. At this point we basically don't need any load line calibration anymore, as the CPU knows which voltage it wants and which voltage it receives, so it can simply adjust the voltage it requests from the VRMs until the voltage received matches the voltage it requires. LLC is now just there so that the initial voltage falls in the right ballpark before the CPU can adjust it. The system might crash if the CPU increases clockspeed and the voltage it receives is too low before it can adjust it upwards. In some monitoring tools you can actually adjust the frequency of polling the data, so you can see smaller spikes as well. The CPU itself should always have a very accurate reading which is constantly updating. If that doesn't help, you'd need to wire an oscilloscope to the VRM.
@@razorblade7108 "golden samples" usually have a higher lifespan. no data for it but its expected to have a longer one in a normal use case (no oc) which it is for this cpu & barely anyone knows how "good" their sample is. since it prob doesnt spike as high. & those spikes could also happen in normal windows doing nothing besides windows background tasks. just let windows defender doing his weekly scan cuz of idle & the cpu will spike maybe not as high as for cpu rendering but high enough. i wouldnt give much about cpu temps. like recently amds 9000 chips moved the temp sensor "down" & gives cooler readings now. while the 7000's have their old placement. idk whats more accurate or even important. it way hotter inside the chip in both cases but usually never seen a newer chip die from overheating. it always was some sort of overvoltage
Yep! I had a Q9650 that I overclocked to 3.6 Ghz using 1.43 Volts. It lasted about 6 months and the CPU degraded enough that I couldn't OC it past 3.36Ghz. I eventually had to run it as stock. The voltage being over 1.4 was a giant red flag. I'm glad that I opted to replace my 9900K with a sweet motherboard/12900K combo for $419, when the 13900K launched.
1.4V has been the norm since Haswell and Skylake. My overclocked i7 6700K at 1.4V still doing fine since 2015 (I gave this particular PC to my sister as an everyday study and gaming PC and is still being in use today). I did upgrade to the 5950X and then the 13900K but kept them at default settings because I am too busy doing research on overclocking CPUs anymore.
About the safe voltages and which silicon is better. It doesn't matter who makes the silicon, it matters what are the characteristics of a node and how logic is designed. The fact that we are speaking about the silicon doesn't mean Ohm's and Kirchhoff's law stop working. They don't. Just like you can put a resistor before an LED to limit current (and thus voltage) and make it run on 230V AC, you can put additional logic in the CPU to increase its Vdrop and make it run at a higher voltage. What matters is the Vdrop across the logic. You can make a CPU that would degrade running above 1V and a CPU that will be totally fine at 3V, using the same node. The difficulty is in matching all the routes so the Vdroop across any given transistor stay relatively unchanged no matter the workload put on the CPU. That's also why some voltages, like IMC voltages need to be in certain range vs other voltages. If IMC feeds the logic at point where logic dropped some voltage already, the voltage difference may make logic to crap out. The same applies for discrete designs but to much lesser degree. CPUs are immensely complicated and thus even 0.05V may make a difference. What amazes me regarding whole LGA1700, Intel designed it grounds up, they've put DLVR in the CPUs, much more sophisticated FIVR which allowed their CPUs to convert voltages on the chip itself, feeding different parts of a CPU with correct voltages, independently. And with Alder and Raptor Lake they forgot that? Like what? It truly must have been a rushed design...
I suspect that all 13900 and 14900 cpus are actually only qualified as 13700 but pushed with higher voltage to be able to perform as 13900 and 14900. Now they have to limit the voltage to "fix the bug" that basically reverting the 900 cpus back to 700. If this is the case, then this is a huge scam and should be brought to justice.
I missed the bus on hearing Intel telling MSI to make a 6/2024 Bios update for a 13900k and 14900k to 6.0+Ghz overclock. I'm sure that aged well with a default lite load of 12, and peaking 450Amps....
@@Celatra It's not right, lol. You can spread the load more evenly better on i9. It'll always be waaaay more efficient than i7s at any power limit. It's as if I said 14100=14900k
What were the odds that some silver bullet UEFI BIOS update would not fix there issues overnight? It would seem that 12th gen intel CPUs were pushed enough as is and even more so for 13th and 14th which was way too much. All because Cinebench took over the mainstream youtube tech space.
Cinebench was always a prevalent benchmark, but yeah, the obsession with minmaxing over these completely irrelevant numbers is quite amusing at the best of times. People stream twice a week or rip some DVD's and they are gaslit into thinking they need productivity rigs.
If I am not mistaken... the VDD_CPU voltage is part of the data voltages within the chip, which are related to the VRAM voltage since they cant differ too much. VDD_IMC is one of the Integrated Memory Controle voltages. Its also related to the voltage of the VRAM. Its an interesting theory if its burning out parts of the memory. But that wouldnt explain why dropping core speed would solve that issue.... Dropping VDD cpu and imc should lower your max DRAM speed. intel's x129 code fixes the needlessly high voltage spikes up to 1.6V, now down to 1.55V or so. However, since people have complained about system stability, of course, it ll push more voltage for all the weak silicon chips... I think I would rather be on the x129 microcode with a good undervolt...
Intel might have redefined the industry standard, but VDD used to stand for the general voltage on the drains of the transistors on the chip (which seems to match the way it's used, the other voltages are for specific parts of the chip). The Ds stand for Drain. See also VSS, VCC, VEE, etc.
Ive said here a few times Ive had no issue with my 13700k. But I was curious and updated to the latest bios and micro code avail for me which was 0x125. It ran slower in everything, I had higher cpu core readings and as you said my cpu was running hotter. I have a 420 AIO. Before the update I would idle at around 32-34c. With the 0x125 I was idling around 40c!!! The moment I realized this I went back the previous version I was using and set watt limits and temp limits. Im scared to mess with the volt settings! If they dont get a micro code that fixes this, they should simply refund everyone their money. The product is not as advertised at that point IMO. Someone will sue them! I have a gigabyte Z690 GAMING X DDR4 (rev. 1.0) mobo. Was running version F29 bios, and tried the F30e bios, which sucks! And Im using 3600 DDR4. Has anyone noticed if the type of memory is playing some role in this?
I found out that the newest BIOS with f29d on mine Gigabyte z690 ud ax ddr4 has worse ram stability. F28 let me overclock from ddr4 3600 to ddr4 4000 with same sub timings. While F29d BSOD a few minutes after booted into windows so I need to use XMP. F29D has 0x125 microcode.
I also have an 13700k and updated to that BIOS version. My idle temps are the same as before the update, (32-35 degrees Celsius) perhaps your temps increment is due to some other issue?
Ive used i9 14900KS for 2 weeks and thats true about the issue.. Crash when iam rendering image in 3Dsmax and Corona Render, BSOD after i give him full load task like rendering in 3D, or i cannot open software like Unreal Engine 5 and D5 Render.. it says error at Unreal and Force close at D5 Render.. i don't know what's wrong but i waited for the Intel driver update so i could running some games and work with this CPU.. My SPECS is i9 14900KS, ROG Z790A, i-game RTX 4080 Super, 64Gigs RAM and running at windows 11 pro.. but still randomly crashes, Ive try different RAM, SSD and downgrade to Windows 10 but still no fix on there.. Ive buy the real license with Corona and D5 and still crashes randomly when i give him full load task.. i don't know if Intel have this issue before i build my rig.. come on Intel 🙄
This is why you should always manually undervolt your components and find that power to performance sweet point. It's what I do and even if peak performance is slightly degraded, sustained performance is greatly improved along with stability and longevity. A 5 to 10% performance hit is worth taking in order to double the lifespan of your components.
Makes zero sense. Then just buy the lesser part for way less money. Who even needs to run same cpu for decades, if you want "double the lifespan of your components"? Why?
@@Tsiikki Yeah it's probably one of the better chips. Silicon lottery and all that. But what he's saying is that often you can get away with some undervolting which can lead to a big decrease in power consumption and temperatures while maintaining or only slightly reducing performance since these chips are often pushed harder than they need to be by default. Pushing higher clocks for more performance hits diminishing returns where power starts going up exponentially with largely minimal performance uplifts.
It was my understanding the microcode only fixes the super high voltage spikes when the CPU enters idle, I'm talking in excess of 1.6V for a number of microseconds, don't think you can measure that from any of the motherboard sensors, CPU's should be able to take ~1.4V as long as there are no spikes. There is no point downgrading the BIOS, a microcode update cannot be reverted, its part of the CPU itself and can only be upgraded. The CPU will ignore any request to update to a lower version. As for the cause, it was my understanding the voltage spikes are destroying the ring bus as the voltage is directly linked with the CPU core voltages, but that was only speculation, only intel really knows.
You can switch microcode whenever you want, it's how people disable patches for spectre, meltdown etc. A good example is v3 Xeon E5's, early microcode had a bug that enabled all cores to boost to max turbo, and if you modded this microcode in to the bios and then loaded the updated microcode after booting you got all core max turbo on latest microcode
@@andrev5207 I don't think the ranting is unjustified. 1.4v is still a pretty high voltage to keep something running at as a baseline. The longevity on these chips will probably be dubious even with the transients removed. It shouldn't be unreasonable to pay £400-500 for a cpu alone and expect it to last for 5-10 years without being manually overclocked.
If you don't use Intel def settings on a gigabyte motherboard with this new microcode, your 1.55v limit will be disabled... so it's pure shit. ua-cam.com/video/TOvJAHhQKZg/v-deo.html because with intel def it request too much voltage compared to a custom config with IA voltage limit.
It took one month for my 14900KS to degrade and that is with the intel default settings(spec sheet) and manual clamping on IA DC/AC loadline configs, PL1/PL2 and IA VR after finding it was still going too high. I got a new cpu from warranty but I'm not going to put it in until there's an actual fix. Right now I'm using a 7800X3D.
@@hank_bloodshadow I ask this because lately I noticed some errors like crashes kernel 41. I got this 3 days ago just after reinstalling Windows thinking it would probably be an update error while I was playing
I noticed sometimes i get explorer to bug out, hangs or even the taskbar disappears sometimes. I have a 13900k with the microcode update. Is it possibly a degradation?
The reason I bought a K variant was especially to have all control on voltage & turbo so I could lower it on day 1. 5.2GHz stable at 1.19V VDD. 1.23V VID dropping with power draw, tuning load line calibration makes a lot of difference on total power draw. Also forget about insane single core frequencies those are the worst in terms of voltage. Lock the turbo clock accross the board all cores same as 1-2 cores. it's losing 2% performance but we knew with Intel only pushing frequencies higher that this was not a good solution.
Where is this safety spec sheet you have ? Because I cant find anything on the voltage from Intel website, also the problem was never the height of the voltage but the travel to there from what I understood from Intel's explanation of the problem.
I'm that guy that wanted to buy a 14700F and have it work, out of the box, with no tinkering. I don't know, nor do I want to know about VID's and clocks. I needed a company who I could rely on but apparently making fast and reliable CPU's is just a secondary activity for intel - primary activity is money extraction...
I'm gonna take Buildzoid's word for it over TYC 10/10 times. He actually knows what all the voltages do, what should be safe, and actually measures the voltages with a proper tools instead of trying to rely on hwinfo.
It doesn't help when motherboard vendors use different names in the voltage menu try find vdd_cpu voltage in the gigabyte menu 😐 there's no such thing 2:52
I have the 0x129 microcode beta update on my MSI z690 MPG Wifi. What do I do to manually limit my voltage that it doesn't hit the limit in BIOS voltage?
TSMC silicon is just superior. People were freaking out about Zen 4 hitting 95c, when Apple lets their TSMC silicon go as high as 106c. AMD over corrected with Zen5 basically to shut people up, now they're complaining that it's too held back. ECO mode was always an option that retained your warranty, and now you have to violate it to get the performance that would've come if people didn't whine so much.
No one beats physics. even if your CPU can go over 95C it's still not good for it and will break over time. Apple doesn't care about long lasting products.
Undervolting etc also voids warranties. The warranty line is just boilerplate to cover their arse so AMD/Nvidia/Intel/MSI/Asus/Gigabyte/Razer.... can deny you while pointing to it and say sorry. You have as much warranty on anything you buy and its luck of the draw, so you do what you want with it and hope for the best.
The micro code is garbage. It’s not a fix it’s a bandage meant to last long enough for people to be out of their warranty. To prevent this once you get an Intel I9 or I7 just sync all cores and set clock speed to 5.7Ghz. Also put on a 253 watt power limit. The reason why this is happening is because 1.5 volts is being sent to 2 of the cores. It’s also a fast voltage spike as well. This is all preventable though.
Does your bios not have Intel recommended profile (the motherboard manufacturer bios profiles disables the 129 patch voltage limits, from tests I have seen)
I got 11900K and 10900K on clearance and the Z490 motherboards were extremely cheap because no one wanted them after Alder Lake came out. It's also the last chipset that you can make a hackintosh easily as it's Apple's last Intel platform, so I'm using 11900K as my workstation PC and I'm building a Hackintosh around 10900K. I decided to wait it out until DDR5 got cheaper (I also got Ryzen 5700G for mom) so I dodged so many bullets.
Also, 11900K's AVX-512 is faster than Zen4/5, and a lot of emulators are taking advantage of them now. You can make some really boutique specialty builds with 10900K/11900K Hackintosh/PS3 Emulator
Most people back then suggested the 10th gen with the b560 and screwed people out of their top M.2 slot. All for price to performance. Discord was loaded down with that "Upgrade path" back in 2021.
I don't disagree with you about people still having issues after this microcode. But I think what you're seeing here is a seperate issue. This microcode is supposed to eliminate voltage requests above like 1.54v during light load spikes. Buildziod did a vid on it
What a waste of time this video was. Especially at the end where he slags off Intel vs TSMC. Completely wild speculation from someone who has done no real physical voltage testing and has no idea what they’re talking about. Buildzoid is who you should watch on this for real insight.
Yea sure buddy. But my CPU is still drawing +1.6v with the new BIOS on defaults. TECH Yes City is right. On paper it should not. But real world tests prove that the voltage is still too high which will still cause degradation.
The VDD Setting is configured by the Motherboard manufacturer, Intel has clear specifications for that (at least in this one setting), maybe because is a Jedec standard. It's been super high allways since 6th gen Intel. I allways tune it down. Edit: If you want to lower manually you can set it to 1.25v (That should suffice up till 6800MT/s)
Perhaps the best assessment yet on the Intel issue. Would love to know your thoughts about the difference claimed between the Desktop & Laptop versions of the i9-14900K(desktop) & i-9 14900HX (laptop). Are they so completely different that laptops don't have the problem? Or are laptop users just not as savvy at detecting these issues? Seems like Intel wouldn't re-invent the wheel that much between them, and likely uses the same silicon, correct? Asking because I just bought an Intel laptop & it's thermally throttling despite Extreme Performance fan settings but can't be undervolted. Still have 116 microcode from BIOS IIRC. Definitely not 129. And no, I'm not going to re-paste the chip on a just received, out of the box, laptop from a major brand.
Seeing the wattage and temperatures between updates seems so insane.. Honestly why I've preferred eco-mode and undervolting. Eg, I run my 5950x at 65watt, it runs quiet and cool, and is still a great performer for productivity workloads. Completely agree, bring back the headroom, let the enthusiasts overclock, make OCing a product feature again, and bring back CPUs (and GPUs for that matter) to more sane levels of heat and power consumption by default. Great content!
DO NOT FOLLOW THE "ADVICE" IN THIS VIDEO!! - Compared 2022 BIOS to 2024 BIOS, saw voltage increase - Clueless about the 2022 BIOS undervolting the CPU by 50-100mV out of the box using loadlines - Thinks Intel is raising voltages using microcode to hide degradation instead of checking said loadlines or the VF table to check what voltage the CPU should have been using - Gives some awful advice about rolling back to old BIOS, setting static Vcore 1.3V, reduce turbo ratio to 52x, and disabling all the E-cores.
I haven't had an issue with my i7 13700k and i use the boost overclocking. I guess it is a hit and miss wit hthese processors but I think the first ones made might have been then issue more than the newer ones
I agree. Really bad suggestions. Nobody needs to disable E-cores. Also 5.2 is really really low. You could just configure Per-core multipliers and configure max multipliers to 57 or 58 for up to 6 cores and 56 or 57 for up to 8 cores. Then try settings -100 or -150mv for all VF points for 51x and above. And set power limit 1 and 2 to 253W.
To be fair, when I built my Ryzen 5600 system on an Asus motherboard, auto overclocking was on by default and VDD jumped crazy high on load, up to 1.47V. When I turned PBO curve optimiser to -15, the VDD jumps to 1.18V max while the processor boosts to 4,65 Ghz (thats 150 Mhz higher than advertised).
How does disabling E-cores help if the problem, as you say in the video, is P-core related? I have an i5 13600K. Should I disable them too? I know they don't help much in gaming, but they speed Windows noticeably.
Spreading misinformation isn't going to help things, mate. 1.424v under light load is NOT "out of spec". The VID tables themselves for almost all 14900K go above 1.424v. Now if you are seeing 1.424v under full load, then you might have an issue. The reason you are seeing increased voltages is because boards before were undervolting by setting the AC_DC loadline lower than Intel's spec. The voltages you are seeing is not what was killing the chips, it was the chips hitting 1.6v+ spikes when transitioning from heavy load to light load.
If you can link me the tables for safe voltages specs that would be appreciated. Other than that I will stick to motherboard guidance, which historically has been accurate.
Also e cores enabled sure. Vdd spikes really high. Though I'm comparing what these new updates do for specifically p cores, and the vid voltages that are assigned.
Pl1 and 2 limits are wattage limits? Not sure what that has to do with voltage safety limits. You can't bust a cpu with a voltage cap in place and a pl1 limit of 4096. Though you can bust a cpu with a high voltage and low wattage limit in place.
@@techyescity It's just to say that a mobo showing something in red or not has nothing to do with safe settings, 4096W has never been safe but no mobo has ever shown it in red.
Really nice information here. I have a question. 13600k stock(almost) is running at 1.48 Vcore in desktop. In game its like 1.3-1.35 Vcore. The only thing i did to my cpu is SA Voltage to 1.33 for my ddr4 ram oc. I tested this for stability like 24/7 and think the ram is really not a problem at all. But heres my point. Even if i try llc , lite load, voltage offset and any of these combined together im getting game crashes every few minutes. Its crazy. Even only reducing the voltage by 30 mV. Do you guys think the SA V is maybe causing me this trouble? I would actually set my ram and SA to stock if this could solve this more or less. Maybe im just going to disable the e cores and try everthing off that point and leave it at that. The thing is im not quite getting pc crashes yet and i really dont want to bother with warranty return, where i wait 2 weeks for my cpu to come back, only because "there's no issue found".
Yup, that sounds likely why they limited spikes but increased the overall voltage to try and stabilise already degraded CPUs. Slower for hotter, pretty crap.
Hello, I have an i5-14600kf, I put the vcore at 1.2v and individually left the limit on each pcore at 5ghz, the question is, am I keeping everything very low at 5ghz or could I keep it at the factory default at 5.3ghz, remembering that I'm using the latest bios 2503 on a rog strix z790-h, the bios configuration is in the Intel standard without touching other things, I just activated XMP and these configurations already mentioned. (Note: I feel the machine is extremely stable and the cpu is new, used for the first time just now)
Very well done, Extremely well done infact. My hats goes off to you. Some UA-camrs are tweaking their sh*t for hours and pretend they don't which is a bit disingenious to me. But you stay true to show us what exact settings trigger what behavior. You don't have a biased fanbase which probably helps. Regardless of stupid logo's we are just trying to survive anyways. Thank you dude! Cheers.
If you read intel statement ; The issue was identified in late 2022, and with the manufacturing improvements and additional screens implemented Intel was able to confirm full removal of impacted processors in our supply chain by early 2024. However, on-shelf inventory may have persisted into early 2024 as a result. I got a new batch of 14700kf and i watched its behavior vddq is locked at 1.4v and v.core 1.3v, it is uv -0.95 and Oc 5.730 ghz p core and 4,4 e cores.
It's been also reported that disabling the Intel baseline profile in the newest microcode bioses removes the new voltage limits and the CPUs can spike up to 1,6 volts on the vcore. This architecture has failed to deliver.
Fully overclockable CPU...if you don't follow intel specs you can fully overclock it, meaning you can use settings that can hurt the CPU. The whole idea of intel defaults is to make sure that users have a way of knowing that mobo makers use or don't use safe settings.
@@dankvader420 It's not the mainboard manufacturers that are at fault for removing the voltage limits when disabling intel baseline?! Whos fault are bios settings then?!
@@Celatra I have not seen a single report of a fried intel CPU (13-14th gen) , maybe you are thinking of ryzen 7000 that would blow up, that you could call frying , for intel they "only" degrade, because they ask for too much vcore which these microcodes are trying to fix.
I think I was lucky when I got 13700kf almost 2 years ago. I followed skatterbencher overclocking video I locked cpu voltage to 1.4v. I think this saved my cpu from degrading cause I did try the default voltage but when I saw the vcore around 1.45 in hwmonitor I immediately locked it to 1.4v. I never experienced any blue-screen or anything. This taught me to always control my system rather the microcode runs it wild.
@@kramnull8962 isn't 31-32k pretty normal for 13700k? Unless you meant that you have to use power saving mode. I really hope for you that Intel will improve their rma process, because having to do multiple attempts when a single one usually means that people are stuck without pc for several weeks is insane.
Vdd is the main voltage fed to the Drain of the MOSFET transistors. In the case of TTL circuits, which use bipolar transistors instead of MOSFET, the general power supply voltage is called Vcc, where c stands for Collector.
AMD released their 9000 series CPU leaving overclocking to the consumer and *most reviewers are grilling them* so I don't think we are going back to releasing CPU not overclocked from the factory any time soon. AMD's marketing team didn't help their case with their promotion materials misleading expectations so it is what it is =S
If AMD told consumers ahead of time that it was basically an efficiency upgrade, people would be fine. I think they are great. I was not impressed with the Ryzen 7000X chips. They run too hot and consume too much power.
When it’s an artificial benchmark race selling modern CPUs it’s a race to no where without any major IPC architectural changes .. bode shrinking hasn’t gained a lot of ipc since tsmc 7nm it has only gotten more efficient and more dense … hence the rtx4000 generation. Ryzen 5000 is still very competitive with 7000 and 9000 … they have already reached the limit of the technology at this point they are just refining the process generationally.. with the cpu side the only way to win for day one reviews is to push voltage to hit targets to be at the top of the reviewers cinebench runs haha… it’s kinda stupid but if Intel would of just locked all cores instead of hand the single core boost algorithm to hit the 6.0ghz for single core benchmarks it’s burning up those cores pushing it that hard should of just left it at 5.7 all core and said fuck it if we lose some artificial benchmarks we lose some this gen … but the way incompetence has spread in every industry in the country it wouldn’t surprise me at all if Intel only has the most incompetent of leadership at this point they are so much bigger than AMD who I’m sure are just more efficient and competent at this point where it matters but Intel isn’t being blown out the water really they are tied basically with a lil worse power draw last gen not that big of a deal AMD survived a decade being worse performance per watt haha they should of just buy the bullet and worked on there next generational products instead of refreshing with higher power draw algorithms..
Gamers think AMD exists to serve them. Their current focus is on datacentre. They make datacentre cores , and sell some to retail. To use an analogy, think of Zen4 as the standard pentium, and Zen5 as the Pentium MMX. The p166 and p166mmx were identical in most software. It wasn't until software was compiled for MMX that it pulled ahead. AVX512 is basically the same as MMX. It is a SIMD unit. It's wider, and floating point, but the general principle is the same. So AMD are selling "Ryzen MMX" Datacentre folk love it. Gamers don't see the point. In time, they probably will.
@@Austin1990 Same boat here. I will wait for the usual "AMD price drop" as I run some servers at home so I would really appreciate the efficiency gains but not at launch prices =)
Wow, this is horrible advice. For one, youre forgetting and not testing for transient voltage spikes. The previous versions during single core loads were spiking as high as 1.6v, THAT is what causes degradation. The new updates remove the transient spikes. And as for your advice, you dont need to gimp your chip like that. To eliminate the spikes on single core loads, lock your core multipliers to the default boost clock for your chip. This prevents the transient spikes under single core load because it doesnt try to over boost those preferred cores. Jayz2Cents confirmed with one of the SI's that the transient spikes were on single core workloads. And you should be able to get your temps down by using a voltage offset or ac load line to eliminate any thermal throttling.
Imagine going through all that to make sure ur 300$ cpu doesnt burn itself down. And while consuming too much wattage for what a cpu should be pulling.
My i7 14700k ran at 1.387v stock in the bios, I reduced the VRM Core Voltage by 0.1050mv taking it down to 1.285, then reduced the Global SVID Core Voltage by 0.100mv. Cpu shows 1.190v in bios. Would you guys think this is great? My cpu benchmarks are the same as stock if not 1% better. Intel utility benchmark does not say Yes to Thermal/Voltage limit (can't remember what it was) and current limit, during the benchmarks where as it did when it was on stock, and with just the vrm undervolt it was just the current limiting during the benchmarks. Problem I believe with the chips is the e cores running on the same rail as the p cores.
Should I update my bios ? I have a new unused i5 13600K i am willing to use, should I use these microcodes and bios updates ? or just undervolt in bios manually, and if yes please let me know how
I wish Intel would just ditch the P core and E core architecture, I know the extra cores look good on paper but it seems to cause so many issues. I also think this is Intel's Bulldozer moment, as bad as it is, AMD's hands aren't clean historically. Hopefully this is the push they need to get back to making better CPU's
E core has never the issue on this problem it all about overvilting from intel on the P core for example my E core can go up to 4.6ghz without issue and running way lower voltage to boot
11900k masterrace. Fastest CPU from Intel that won't randomly kill itself and support AVX512 stock (yes, you can disable E-cores on socket 1700, but that kinda defeats most of its appeal).
12th gens have no issues and AMD 7k series is good from AMD, newer 9k i would avoid till prices drop as it isn't good at those prices vs the 7k series and all reviews show the same thing unless you want to sacrifice all that efficiency for PBO output and that is a lottery to a great extent, but those results can be sen at Jayztwocents YT page currently. I wouldn't be looking backward technology wise but you are free to buy what you like. Cheers!
Hey I got a i9 13900kf have it for over a year now, never updated the bios, got a msi mag z790 tomahawk wifi ddr5, its oc to 5.2ghz all core with undervolt offset - 0.110 and its never pushing more than 1.275v and 200w, max temps in cinebench 86c with score 38k, dont have any stability issues never get blue screen, only time I did get a blue screen was when I pushed the undervolt to high like 0.150v other than that no issues, the question is do I have to do anything when there are clearly no issues and don't really see the point?
Um hate to break it to you but without an oscilloscope+ counter, the only way to actually see micro code working or not ( Hint its already out and on YT now ). Not going to do that with HWinfo etc apps. I used to do this stuff for a living, Apple, IBM, MS, Game Studios etc. retried hardware eng. Lets set the proper example for folks if you don't mind. While HWinfo is a great utility it just is not designed to do this specific job. We know we been doing this for much longer than a few years =)
Was happy to see someone else said it. The sensors in hwinfo is a good guide, but it's not the real picture. I normally like Tech Yes City, but he's downright wrong here. Intel set a hard limit on the spikes to stop them going above 1.55v. The 1.424v under light load is NOT out of spec, it's literally part of the VID table. 1.424v is actually low compared to what I expected him to get mad about.
@@techyescity ua-cam.com/video/SMballFEmhs/v-deo.html educate yourself sir. Not an issue of faulty sensors, they are not capable of the fine increments and speeds of those measurements. You simply can't see nor measure everything properly without one and a counter. By all means don't believe me ask any hardware engineer or Fab builder. Cheers!
I disabled my e cores except for 4 of them. Now when I download files from a 3d website, it takes forever to download them. Is that because I disabled the e-cores? Thanks
From what I've seen it helps the peak vid table draws but they can't lower it enough because it will eat into performance. That has to happen to fix the degradation and make it stable. Now with this microcode it might help them live longer but it will not fix the problem. The problem is fixed by disabling e-cores and putting the CPU to somewhere between 5.4-5.7ghz. Along with limiting the peak power that the CPU can even request (check Actually Hardcore Overclocking for a video on that). Long story short, the raptor lake P-cores need to be downclocked to 5.4-5.7 for it to be stable and not shove massive voltage into the core. End of story.
Are laptop processors also experiencing elevated voltage issues? I have an i5-13420H with max TDP of 45W. Should I be concerned about a microcode update?
No, it’s dumber than that. There is a bug in the code that disables voltage limits so you got it going over the safe amount. It’s not a thought out thing, just a rushed oversight
I have the ASRock PG Lightning /D4 board (I think you have the D5 version in this video). I found that setting the CPU Core Current Limit to 310 on my board keeps the speed about the same as prior to the microcode update and keeps the temps down and the VDD CPU never going over 1.336. I also had to disable my RAM OC to keep the VCCSA voltage at stock.. For reference I am using the cheapest 280mm water cooler I could find and it holds at 95c under the Cinebench r23 load... Am I fooling myself, or is this a safe(ish) setting to use...?
Why are you manually changing the Voltage? The whole point is to not do that. Use the default Intel settings. To me this only proofs "somebody" has OCD and just cannot help changing things
Hello @Tech YES City and other viewers .. I have a Gigabyte Z790 UD AX board updated to latest BIOS F11e with 0x129 fix running a borrowed i5 12th gen. I have to upgrade my processor. I wanted to go with i7 14700k but with all these issues going on with instability and processors degrading; do you think it is advisable to go with an older 12th gen CPU like maybe an i9 or i7 of 12th gen and compromise using an older generation instead of opting i7 14th gen to avoid this mess ? what is your opinion in my case. Thanks
He's seeing increased voltages because the Intel Baseline Profile brings the AC Load Line back up to their spec which causes less vdroop under heavy load. While this gives higher overall voltages, it also helps lessen the extreme voltage spikes. Nothing stopping you from putting a negative offset on your core voltage if your CPU can handle it, but a lot of CPU's, even brand new, would be unstable under the old load line settings.
@@ZoneXV if so, the intel baseline profiles have been around a while, and he conflated the microcode update with a bios profile change. Again, poorly controlled bios configuration invalidates the premise. You test all conditions equal, not unequal.
Stumble at best..This is Intel they are not going anywhere. They attempted to produce their own chips first time instead of all TSMC produced and it didn't turn out great to be honest hence oxidation issues they experienced in fabrication over 13 sku lines. This is why some people are having no issues at all while others suffer greatly.
@@tysoncoffman7562what are you on about? Intel has always fabbed the majority if not all of their CPUs in house, part of their problem is how far they have fallen behind TSMC vis a vis fab processes
@@purplepioneer5644 No they haven't. They in fact make fab themselves in such places as Oregon, Arizona, Ireland, Israel, Design Fab in Folsom Ca, Malaysia as well as other places and contracted work from/thru TSMC in fact they just recently spent 14 billion on manufacturing it new chips. Intel's Mix- Source strategy. Like it or not its a fact. Do go look it up, and educate yourself. Cheers!
Does anyone know which values you have to change on an asus bios, since cpu_vdd voltage is named differently, also for the memory controller but that is a little more eyplanatory
Intel doesn't want you to know the permanent fix.
But it's Tech YES Lovin.
Because there's none
@@sebastianandres6802 Did you read my comment?
Well said Marco!
Yes! Indeed.😊
@@sebastianandres6802 No there is a permanent fix. The bulk of their CPUs can't actually handle having boost clocks at 5.8-6.0Ghz, because the voltage required is too high to not cause degradation. Max boost should probably be 5.5-5.7, which is too equivalent to AMD to have a competitive advantage.
Permanent fix is replace with X58 and cheap Xeon.
I think their biggest lie is stating flat out that the previous microcode was somehow acting in error, or giving "incorrect" requests - it was doing exactly what they wanted it to. Their motivations were purely to beat the 7950X in benchmarks and advertise a boost clock of 6Ghz, and they tweaked everything they could specifically to squeeze every last drop of blood from a stone essentially.
If I got that right, then the standby voltage was too high as well. This is not in line with your assumption that they did it on purpose.
Still if they did it on purpose like you said, they made an unintentional mistake by giving to much standby voltage.
@@WCIIIReiniger nope. they literally tested and shipped these cpu's with those settings. intel has been desperate trying to beat amd by brute force.
@@WCIIIReiniger
I don't think idle voltage has increased since 0x125.
@@WCIIIReinigerI get the feeling someone hired you to say that lol
@@zCaptainz I just stated what I heard in a different video, not sure if it was Tech Yes City, Hardware Unboxed, Gamer's Nexus - probably it was der8auer.
I don't own any Intel product and don't care about their economic performance.
If Intel was a body shop, they would paint over rust. I'm an ICT architect and my brain doesn't fathom Intel silicon being non-compliant with manufacturer specification when they are the manufacturer
They didnt validate the design thoroughly enough, so they kinda don't know what the actual spec is
That analogy of painting over rust is so true in this regard lol...
They rushed the product (Raptor Lake, 13th gen) out in 11 months when most processor arch iterations took 3 to 6 years. Unsurprising :\
You misses the part where Tech YES City got it all wrong. Microcode works correctly on undamaged chips only. Pulling out a CPU that was acting up for testing was pointless.
@@Reeper-x6x Yeah lets fix a broken CPU, that's cooked.. it runs, but its core phreaked, over peaked n wouldn't even be worth throwing on eBay for the grief...
"it's a massive chicanery of fuckery" --Tech YES City
I was having my lunch at the time, wasn't even looking at the screen and almost spilled everything out, could not stop laughing. I had to rewind to listen to it again. Let the class action lawsuits begin😂
He defecated through a sunroof!
@@gingerbread6967 The possibility for a class action lawsuit is already being investigated. Intel has known the issue for at least 2 years and it is not difficult to prove it in court. Tech Yes City is not ''misinformed'' alone, Level 1 Tech, Actually Hardcore Overclocking, Gamers Nexus, Hardware Unboxed, Jayz 2 Cents, and a few others are also on the ''misinformation'' train.
Lots of ''misinformation'' going around these days.
@@gingerbread6967 explain
Ugh, what a huge mess Intel made. I still can’t believe Intel didn’t stop selling these CPU’s after admitting there’s a problem
They even doubled down on these issues by releasing the second flawed generation. I dont think someone with rational thoughts is making decisions at intel. Since sooner or later it will cut into their own flesh. And the damage to the brand is indeed massive. No one in their right mind would ever by Intel again after going through this mess.
Intel after this microcode: Whoops wrong code.
😂
That made me think of Cheech and Chong, up in smoke. He gives Cheech the wrong pills.
And this made me think of that sketch saying "oh, different Abby". 😂
It's remind me recent CrowdStike case.
Whoops! wrong update patches~
Stop rushing things. You gotta blame on the motherboard first
They should learn to code.
I remember when it was the end user who used to overclock, now we got the main manufacturer overclocking (to the point of degradtion) and it is the end user who is now dialing it back... strange world we live in.
What's crazy is that they don't even do a good job of it. My Ryzen 7600 out of the box with a big air cooler and PBO enabled used to run at 85c and couldn't hold full boost frequency on all cores at the same time. Then, I let the Ryzen Master curve optimizer do it's thing. Now it runs at 65C and holds full boost on all cores at the same time. It gained like 300-400 CB points too. Either I won the silicon lottery or PBO just does a horrible job with stock settings.
@@madmatt2024 For what it is, PBO is incredibly useful for how little effort it takes, even if there could be better settings. it's not supposed to automatically fine tune for your CPU, but it allows for you to experiment and do that.
Overclocking is not the sole cause of the degradation.
PBO is the worst implementation cause they want it to age the cpu more faster so instead of lasting a long while it dies prematurely. On account of that every1 should disable pbo. I found an article that benchmarked with pbo off and it really doesnt gain the cpu almost any benefit to have pbo while the cost is like 20c extra temps and downclocking during load
Exactly right though! Using Air cooling, just to enhance the problem.
This is why you don't want chips implanted in your body.
I mean, not really. Of all the reasons I don't want chips in my body power draw isn't even a factor. Your brain runs on about 15w, I'd imagine there's not much energy to spare lol
@@Wobbothe3rd😹
@@Wobbothe3rdI hope they install some fan headers so I can go for a nice RGB setup with my brain chip
and the heat...
LOL - Well you know how it goes, you'll have no choice to accept them at some point or get "eliminated", as per Mark Of The Beast :P
13700K, disabled e-cores early on, no issues except a random crash here and there playing Apex (but that could also just be an Apex thing), and that latency on opening certain folders and video editing that I've been complaining about across a couple of Intel CPUs now (13700K and 10980XE). The PC is almost 2 years old now, so... should I be expecting issues later or is it possibly ok?
Wait, Skylake X also has these latency issues?
Don't update your bios for now, and I'll help you tune it when I'm back in aus brother.
10980xe wasn't anywhere as responsive as a 10900k. I think rocket jumps most stable system atm is a 10th gen i9 10850k or 10900k?
Yeah the 10980XE is actually worse than the 13700K when it comes to latency, don't know why. Once Tech Yes is back in Aus, maybe he can do some tests and figure it out, could make an interesting vid.
And yeah the 10850 and 11900 are my 2 most stable, best performing PCs at the moment. Had I known this was going to be the case, I wouldn't have bought the 13700K thinking I was "upgrading". Going to test the 7800X3D next to see what the overall performance is like compared to the others.
@@RocketJumpNinjaI purchased a 13700k and I RMAD my gpu 3 times thinking it was my gpu , at one point I’m sure GPU was an issue but now I’m starting to think holy crap it was my 13700k all along wow. I think I’m rebuilding my 5900x to test stability with my GPU.
Thanks! Please let me know if my interpretation is correct: The new Micro codes Increase voltages & increase the allowed thermal temps. Those two factors eventually throttle the CPU because of unsafe voltage and and/or unsafe temps. So in the end, the new micro codes are enabling the CPU to reach the originally advertised speeds but degrading the CPU at an exponential rate? right?
Yes if the cpus are degrading then this seems like a way to have them still work temporarily. It's minor in difference but I do expect potentially another microcode update in a few months. I mean there has been actually heaps of them for the 13900k. Some concerning stuff.
@@techyescity I had my PC built by Micro Center because they stand by their customers. I told them how the new Micro Code is the equivelent of Gai Sensei opening the eight inner gates except nowhere near as epic. My 14 series based PC was thermal throttling on their test bench so hopefully they will make it right in some way.
What an embarrassment. The excess voltage was what got them into this situation in the first place.
Intel is done. They chose the worst way to "fix" this.
@@megakedarThis will hurt there imagine it will not bankrupt the company.
I get it. Intel wants to avoid a class action lawsuit of reduced performance on these parts. They increased the voltages to avoid the crashing issues at the expense of chip longevity since they expect not too many people run these 24/7. If a CPU dies under warranty, they would happily replace it until it runs out of warranty
The excess voltage happens on degraded chips, Sadly, a Tech YES City used a CPU he knew was faulty to test. You can't fix a bad CPU, period .It need to be RMA'd
no, u guys understood it wrong. the microcode fixed the voltage spikes that went up to or ABOVE 1.6v & thats what killed the cpus. that happened really quickly u usually dont see unless u have a graph that monitors every tiny voltage change / every ms & thats what the issue was. Intel cant fix already damaged / degraded chips. thats why they offered a 5 year warranty for every chip that was affected by it. (if ur unsure if its damaged u still can make a warranty claim to get a brand new one, or in other words just lie & say its damaged.)
Yes this is probably the legal game.
They have marketed those speeds but not voltages and durability. They will not give a microcode update which includes lower speeds since it opens them to litigation. They'll just juice up the CPU at different intervals to prolong the lifetime until they're out of warranty... gg thanks for playing
If no one complained, Intel will never take action of this. This is why we should be vocal against billion dollar corporations. Thank you TECH YES CITY and other tech UA-camrs for your efforts in the tech community.
"Fake it till you break 'em" - Intel 2024
😂😂😂
I think it's crazy that you pointed this out a year ago and that largely went unnoticed, but I remember watching those. ..Bravo to Mr Tech Yes City. Hats off to you sir
I had zero faith that this code update would solve anything
lol
Im staying on my 10900k for a very long time.
I have 11th Gen I5 and it works fine. I grew up on AMD Intel was feeling like snobby CPU at that time I grew up on. AMD is kicking both Nvidia and Intel butts in the stock market.
You know ryzen exists 😂
I have a 10850K that I don't think I'll be moving from for a long while.
@@LocoMe4u I use AMD, but switching boards and CPU is expensive, especially if he doesn't need to upgrade atm.
@@LocoMe4u tru but there's no point in me upgrading, I run all the games i want at high fps at the moment so I'm happy with it.
This issue is not about voltages UNDER LOAD. Under load, the voltage does not go higher than 1.5 volts. It can't, the chip will power limit to under XXX watts (depending on the chip model). This is not about wattage (power draw). This issue is not about temperature. All people testing under load or looking at temperature to see if the problem is fixed and all the people running motherboard sensor data to determine if the issue is fixed or not are not looking at the cause or the fix.
Firstly, my understanding is that motherboard sensors are not sensitive enough to consistently measure the actual voltage spikes involved. The sensors are sensitive in the milli-second ranges and the spikes in question are in the micro-second range .. so 1000x faster than the senor is designed for. So, the peak spikes shown are completely irrelevant with the actual spike felt by the chip / ring bus.
it is like measuring your blood sugar 5 hours after eating a candy bar. You see and say, my blood sugar is normal. Sure, but it wasn't at 15 minutes or 45 minutes or 120 minutes. But it was great at time 0 and time 300, the only measurements you have.
The motherboard sensors take a reading at time 0 then again at time 1000, then at time 2000 (in a micro-second event). They don't see times 1 to 999, times 1001 to 1999, etc. So, if you get lucky and if the spike actually happened at time 1000 or time 2000, then you see it .. if it happened at a time where you did not measure (a 999/1000, or 99.9%, chance if we are talking a micro-second to milli-second event comparison) then you just don't see that voltage.
Should you just take intel's word for it? Absolutely not! But wait until some people with proper external equipment and knowledge actually use equipment sensitive enough to detect and diagnose the problem actually do it.
Running CPUs at high voltages for a short time does not degrade them a lot, that's why extreme overclocking works at much higher voltages with the CPUs still working fine for years. What degrades a CPU is a combination of continuously high voltage, high temperature and high current. But while current and temperature increase degradation linearly, voltage increases it exponentially. 1.5V is not a safe voltage to run a CPU for a long time. Not even Bulldozer, which was based on a bulletproof 32nm silicon on insulator manufacturing process, could handle more than 1.45V continuously. On current manufacturing processes with much smaller structures, you're better off not going far beyond 1.4V under load or beyond 1.35V continuous if you want the CPU to last beyond the warranty period.
@@razorblade7108 that is true in general. But THIS specific problem is damage not caused by long term high exposer but short term very high spikes. This kind of damage did not exist before now. It is all about how they wanted quick boost ability.
I have no issue running at lower voltages in general. I absolutely undervolt.
But that is not this issue and should not be confused for this issue.
You can limit to 1.4 and you will still get these spikes if you all boosting.
ive seen tests that recorded voltage spikes up to 1.6v & above under load or in the beginning of the load. so its def. a problem. & after the update it didnt spiked to the 1.6v anymore (also core speed lowered by 20mhz or so). the spikes are still relevant since cpu are voltage "sensitive". i agree with the temp & watts but that should be common sense for the most part - still relevant but not in this case.
However not every CPU is a gold sample. every cpu has some "defects", some will run fine at the 1.6v or 1.5v spikes but others cant take those. some clock a bit higher at lower voltage while the next cpu needs 0.05v more for the same speed or it doesnt clock as high as others resulting in 20mhz less or more depending on the chip. + Some motherboards tend to give more voltage than others on top of it. CPU have a margin of errors they can have when they got produced. too many errors/defects disabling certain feature sets/cores = i5 or i3 while the good ones are i7 & i9. some will even work for maybe 5-10 years without the microcode fix without any problem since they prob got the golden sample, others have chips that barely got the i7 mark in production & if u got a chip like that, then gl surviving the spikes long term.
Extreme overclocker using usually a golden sample for those high clock speeds & voltage.
Motherboards sensors arent 1000% accurate but its the best u currently have for the normal customer and for the most part they are accurate enough to notice that there is some issue or change.
@@Vss077 If those spikes are really high and happening often and long enough, they can add to the problem, but in general running any CPU at 1.5V for longer periods of time will degrade it no matter what, even if it's a golden sample. Also a CPU which is a golden sample means it runs really efficient, so it needs low voltage for the same clockspeed, but actually it could also degrade faster at higher voltages. It is less likely to degrade when running at the same clockspeed with lower voltages though.
Regarding the sensors I agree that motherboard sensors are usually not accurate enough, but nowadays you also have sensors all across the CPU, measuring temperatures and voltages more accurately and also enabling them to account for voltage droop between VRM and CPU. At this point we basically don't need any load line calibration anymore, as the CPU knows which voltage it wants and which voltage it receives, so it can simply adjust the voltage it requests from the VRMs until the voltage received matches the voltage it requires. LLC is now just there so that the initial voltage falls in the right ballpark before the CPU can adjust it. The system might crash if the CPU increases clockspeed and the voltage it receives is too low before it can adjust it upwards.
In some monitoring tools you can actually adjust the frequency of polling the data, so you can see smaller spikes as well. The CPU itself should always have a very accurate reading which is constantly updating. If that doesn't help, you'd need to wire an oscilloscope to the VRM.
@@razorblade7108 "golden samples" usually have a higher lifespan. no data for it but its expected to have a longer one in a normal use case (no oc) which it is for this cpu & barely anyone knows how "good" their sample is. since it prob doesnt spike as high. & those spikes could also happen in normal windows doing nothing besides windows background tasks. just let windows defender doing his weekly scan cuz of idle & the cpu will spike maybe not as high as for cpu rendering but high enough. i wouldnt give much about cpu temps. like recently amds 9000 chips moved the temp sensor "down" & gives cooler readings now. while the 7000's have their old placement. idk whats more accurate or even important. it way hotter inside the chip in both cases but usually never seen a newer chip die from overheating. it always was some sort of overvoltage
As a noob overclocker first thing i learned is not to pass 1.4v line yet intel engineers used 1.55v on a cpu sometimes hitting 1.6v.
Just lol.
Yep! I had a Q9650 that I overclocked to 3.6 Ghz using 1.43 Volts. It lasted about 6 months and the CPU degraded enough that I couldn't OC it past 3.36Ghz. I eventually had to run it as stock. The voltage being over 1.4 was a giant red flag. I'm glad that I opted to replace my 9900K with a sweet motherboard/12900K combo for $419, when the 13900K launched.
I nuked a Q after 5 or 6 months also, they didn't like the voltage at all. @@Gamevet
1.4V has been the norm since Haswell and Skylake. My overclocked i7 6700K at 1.4V still doing fine since 2015 (I gave this particular PC to my sister as an everyday study and gaming PC and is still being in use today).
I did upgrade to the 5950X and then the 13900K but kept them at default settings because I am too busy doing research on overclocking CPUs anymore.
sounds like that should be a meme.
Smaller the transitors the lower you can push voltage. It was fine then for that size, but now ~1.2V seems to be equivalent afaik @@paul2609
Intel won't change its behavior until they get served with a multi billion dollar class action lawsuit.
Intel : give them this update rather than we take the chip back....LOL
About the safe voltages and which silicon is better. It doesn't matter who makes the silicon, it matters what are the characteristics of a node and how logic is designed. The fact that we are speaking about the silicon doesn't mean Ohm's and Kirchhoff's law stop working. They don't. Just like you can put a resistor before an LED to limit current (and thus voltage) and make it run on 230V AC, you can put additional logic in the CPU to increase its Vdrop and make it run at a higher voltage. What matters is the Vdrop across the logic. You can make a CPU that would degrade running above 1V and a CPU that will be totally fine at 3V, using the same node. The difficulty is in matching all the routes so the Vdroop across any given transistor stay relatively unchanged no matter the workload put on the CPU.
That's also why some voltages, like IMC voltages need to be in certain range vs other voltages. If IMC feeds the logic at point where logic dropped some voltage already, the voltage difference may make logic to crap out. The same applies for discrete designs but to much lesser degree. CPUs are immensely complicated and thus even 0.05V may make a difference.
What amazes me regarding whole LGA1700, Intel designed it grounds up, they've put DLVR in the CPUs, much more sophisticated FIVR which allowed their CPUs to convert voltages on the chip itself, feeding different parts of a CPU with correct voltages, independently. And with Alder and Raptor Lake they forgot that? Like what? It truly must have been a rushed design...
I suspect that all 13900 and 14900 cpus are actually only qualified as 13700 but pushed with higher voltage to be able to perform as 13900 and 14900. Now they have to limit the voltage to "fix the bug" that basically reverting the 900 cpus back to 700. If this is the case, then this is a huge scam and should be brought to justice.
I missed the bus on hearing Intel telling MSI to make a 6/2024 Bios update for a 13900k and 14900k to 6.0+Ghz overclock. I'm sure that aged well with a default lite load of 12, and peaking 450Amps....
they do have a bit more cores and more cache, but that's about it, architecturally they are the same, so yeah, you're basically right
@@Celatra
It's not right, lol.
You can spread the load more evenly better on i9. It'll always be waaaay more efficient than i7s at any power limit. It's as if I said 14100=14900k
@@tteqhu if that were true, the I9 wouldnt have to consume 400+ watts just to have a 5% lead over a i7
@@Celatra
Where are you getting those numbers from?
What were the odds that some silver bullet UEFI BIOS update would not fix there issues overnight?
It would seem that 12th gen intel CPUs were pushed enough as is and even more so for 13th and 14th which was way too much.
All because Cinebench took over the mainstream youtube tech space.
Cinebench was always a prevalent benchmark, but yeah, the obsession with minmaxing over these completely irrelevant numbers is quite amusing at the best of times. People stream twice a week or rip some DVD's and they are gaslit into thinking they need productivity rigs.
If I am not mistaken... the VDD_CPU voltage is part of the data voltages within the chip, which are related to the VRAM voltage since they cant differ too much. VDD_IMC is one of the Integrated Memory Controle voltages. Its also related to the voltage of the VRAM.
Its an interesting theory if its burning out parts of the memory.
But that wouldnt explain why dropping core speed would solve that issue.... Dropping VDD cpu and imc should lower your max DRAM speed.
intel's x129 code fixes the needlessly high voltage spikes up to 1.6V, now down to 1.55V or so. However, since people have complained about system stability, of course, it ll push more voltage for all the weak silicon chips...
I think I would rather be on the x129 microcode with a good undervolt...
Intel might have redefined the industry standard, but VDD used to stand for the general voltage on the drains of the transistors on the chip (which seems to match the way it's used, the other voltages are for specific parts of the chip). The Ds stand for Drain. See also VSS, VCC, VEE, etc.
Me watching this whole saga unfold with my 5600x 😎.
Same but with my 7950x3d
Replace x with H and that's me
That’s a budget CPU
GT version is here😎
And still gaming at 60-100fps in 2024🤣
Ive said here a few times Ive had no issue with my 13700k. But I was curious and updated to the latest bios and micro code avail for me which was 0x125. It ran slower in everything, I had higher cpu core readings and as you said my cpu was running hotter. I have a 420 AIO. Before the update I would idle at around 32-34c. With the 0x125 I was idling around 40c!!! The moment I realized this I went back the previous version I was using and set watt limits and temp limits. Im scared to mess with the volt settings!
If they dont get a micro code that fixes this, they should simply refund everyone their money. The product is not as advertised at that point IMO. Someone will sue them!
I have a gigabyte Z690 GAMING X DDR4 (rev. 1.0) mobo. Was running version F29 bios, and tried the F30e bios, which sucks!
And Im using 3600 DDR4. Has anyone noticed if the type of memory is playing some role in this?
I found out that the newest BIOS with f29d on mine Gigabyte z690 ud ax ddr4 has worse ram stability. F28 let me overclock from ddr4 3600 to ddr4 4000 with same sub timings. While F29d BSOD a few minutes after booted into windows so I need to use XMP.
F29D has 0x125 microcode.
I also have an 13700k and updated to that BIOS version. My idle temps are the same as before the update, (32-35 degrees Celsius) perhaps your temps increment is due to some other issue?
Ive used i9 14900KS for 2 weeks and thats true about the issue.. Crash when iam rendering image in 3Dsmax and Corona Render, BSOD after i give him full load task like rendering in 3D, or i cannot open software like Unreal Engine 5 and D5 Render.. it says error at Unreal and Force close at D5 Render.. i don't know what's wrong but i waited for the Intel driver update so i could running some games and work with this CPU.. My SPECS is i9 14900KS, ROG Z790A, i-game RTX 4080 Super, 64Gigs RAM and running at windows 11 pro.. but still randomly crashes, Ive try different RAM, SSD and downgrade to Windows 10 but still no fix on there.. Ive buy the real license with Corona and D5 and still crashes randomly when i give him full load task.. i don't know if Intel have this issue before i build my rig.. come on Intel 🙄
@@j_shelby_damnwird Not sure what it could have been The room was the usual 72F. I went back to F29 and the idle temps went back to normal. Not sure.
they are getting sued though???
It's hilarious, they got customers and shareholders sueing them in two separate lawsuits.
This is why you should always manually undervolt your components and find that power to performance sweet point. It's what I do and even if peak performance is slightly degraded, sustained performance is greatly improved along with stability and longevity. A 5 to 10% performance hit is worth taking in order to double the lifespan of your components.
Makes zero sense. Then just buy the lesser part for way less money. Who even needs to run same cpu for decades, if you want "double the lifespan of your components"? Why?
I have a manual PBO tune on my 5800x3d, it actually performs better than stock with lower power draw.
@@Son37Lumiere well there's that. But there are probably parts that need the spec voltage. So GG for better than spec CPU!
@@Tsiikki Yeah it's probably one of the better chips. Silicon lottery and all that. But what he's saying is that often you can get away with some undervolting which can lead to a big decrease in power consumption and temperatures while maintaining or only slightly reducing performance since these chips are often pushed harder than they need to be by default. Pushing higher clocks for more performance hits diminishing returns where power starts going up exponentially with largely minimal performance uplifts.
@@Tsiikki Always undervolt.
I have a 12th Gen i9. Would updating to the new bios have any bad effects on mine?
If all is good ,don't upgrade.
@@snoopdodoO good luck
It was my understanding the microcode only fixes the super high voltage spikes when the CPU enters idle, I'm talking in excess of 1.6V for a number of microseconds, don't think you can measure that from any of the motherboard sensors, CPU's should be able to take ~1.4V as long as there are no spikes.
There is no point downgrading the BIOS, a microcode update cannot be reverted, its part of the CPU itself and can only be upgraded. The CPU will ignore any request to update to a lower version.
As for the cause, it was my understanding the voltage spikes are destroying the ring bus as the voltage is directly linked with the CPU core voltages, but that was only speculation, only intel really knows.
The microcode is loaded into volatile storage on the CPU at each boot, so it can be ‘reverted’.
Finally someone actually understands 👍. In the meanwhile everyone else keep ranting even when problem is fixed
You can switch microcode whenever you want, it's how people disable patches for spectre, meltdown etc. A good example is v3 Xeon E5's, early microcode had a bug that enabled all cores to boost to max turbo, and if you modded this microcode in to the bios and then loaded the updated microcode after booting you got all core max turbo on latest microcode
@@andrev5207 I don't think the ranting is unjustified. 1.4v is still a pretty high voltage to keep something running at as a baseline. The longevity on these chips will probably be dubious even with the transients removed. It shouldn't be unreasonable to pay £400-500 for a cpu alone and expect it to last for 5-10 years without being manually overclocked.
If you don't use Intel def settings on a gigabyte motherboard with this new microcode, your 1.55v limit will be disabled... so it's pure shit. ua-cam.com/video/TOvJAHhQKZg/v-deo.html because with intel def it request too much voltage compared to a custom config with IA voltage limit.
It took one month for my 14900KS to degrade and that is with the intel default settings(spec sheet) and manual clamping on IA DC/AC loadline configs, PL1/PL2 and IA VR after finding it was still going too high. I got a new cpu from warranty but I'm not going to put it in until there's an actual fix. Right now I'm using a 7800X3D.
How did you notice the first signs that your CPU was degraded?
@@MrAntastik Things that used to be stable became unstable, regular bluescreens and complete shutdowns and automatically restarted itself.
@@hank_bloodshadow I ask this because lately I noticed some errors like crashes kernel 41. I got this 3 days ago just after reinstalling Windows thinking it would probably be an update error while I was playing
I noticed sometimes i get explorer to bug out, hangs or even the taskbar disappears sometimes. I have a 13900k with the microcode update. Is it possibly a degradation?
The reason I bought a K variant was especially to have all control on voltage & turbo so I could lower it on day 1.
5.2GHz stable at 1.19V VDD. 1.23V VID dropping with power draw, tuning load line calibration makes a lot of difference on total power draw.
Also forget about insane single core frequencies those are the worst in terms of voltage.
Lock the turbo clock accross the board all cores same as 1-2 cores.
it's losing 2% performance but we knew with Intel only pushing frequencies higher that this was not a good solution.
Where is this safety spec sheet you have ? Because I cant find anything on the voltage from Intel website, also the problem was never the height of the voltage but the travel to there from what I understood from Intel's explanation of the problem.
I'm that guy that wanted to buy a 14700F and have it work, out of the box, with no tinkering. I don't know, nor do I want to know about VID's and clocks. I needed a company who I could rely on but apparently making fast and reliable CPU's is just a secondary activity for intel - primary activity is money extraction...
Now this is getting confusing, Actually Hardcore Overclocking says this new microcode delivers safer voltage now.
buildzoid is right. he is doing good testing.
0x125 has given me a performance Increase on my Asus Dark Hero…
I'm gonna take Buildzoid's word for it over TYC 10/10 times. He actually knows what all the voltages do, what should be safe, and actually measures the voltages with a proper tools instead of trying to rely on hwinfo.
@@ZoneXV My thoughts exaclty. 50/50 though, still waiting for Gamers Nexus' take on this.
@@HeartlessVin There is a reason Steve uses Buildzoid as a source :D
This is so horrifying. What a nightmare for people. Thank you for getting this out so quickly.
Great video. Was in the dark before you. Keep us posted please. 13900k owner here and Im scared to turn on my pc...
It doesn't help when motherboard vendors use different names in the voltage menu try find vdd_cpu voltage in the gigabyte menu 😐 there's no such thing 2:52
Darn. I wish you could run through these options for Z790 on HP Omen. The labeling in the bios is different :(
6:39 tell me your game is unoptimised without telling me your game is unoptimised
I have the 0x129 microcode beta update on my MSI z690 MPG Wifi. What do I do to manually limit my voltage that it doesn't hit the limit in BIOS voltage?
at the expense of 15,000 job layoffs 🥺
And some hefty money from the US govt
Every company lays people off.
does this affect i5 13600 cpus? so far, every video seems to stress i9 cores
Applies to every "K" CPUs from 13 and 14th gen from what I seen and I saw a blue screen just today in my 14600k 💀
@@Jin68Jin mine is a kf
TSMC silicon is just superior. People were freaking out about Zen 4 hitting 95c, when Apple lets their TSMC silicon go as high as 106c. AMD over corrected with Zen5 basically to shut people up, now they're complaining that it's too held back. ECO mode was always an option that retained your warranty, and now you have to violate it to get the performance that would've come if people didn't whine so much.
I understand your point but manufacturers need to act like adults and the consumers don't.
No one beats physics. even if your CPU can go over 95C it's still not good for it and will break over time. Apple doesn't care about long lasting products.
Undervolting etc also voids warranties.
The warranty line is just boilerplate to cover their arse so AMD/Nvidia/Intel/MSI/Asus/Gigabyte/Razer.... can deny you while pointing to it and say sorry.
You have as much warranty on anything you buy and its luck of the draw, so you do what you want with it and hope for the best.
I still use a 2012 non-retina MacBook Pro with i7, often hits 100 degrees (CPU and GPU). They don’t make them like they used to 😬
@@matthewtymaja3760 you just lucky, does where the laptops that broke a lot back then, just watch louis rossmann.
Rip to those of us with 14900ks that only have one non-0x129 bios version 😭
The micro code is garbage. It’s not a fix it’s a bandage meant to last long enough for people to be out of their warranty. To prevent this once you get an Intel I9 or I7 just sync all cores and set clock speed to 5.7Ghz. Also put on a 253 watt power limit. The reason why this is happening is because 1.5 volts is being sent to 2 of the cores. It’s also a fast voltage spike as well. This is all preventable though.
Does your bios not have Intel recommended profile (the motherboard manufacturer bios profiles disables the 129 patch voltage limits, from tests I have seen)
I got 11900K and 10900K on clearance and the Z490 motherboards were extremely cheap because no one wanted them after Alder Lake came out. It's also the last chipset that you can make a hackintosh easily as it's Apple's last Intel platform, so I'm using 11900K as my workstation PC and I'm building a Hackintosh around 10900K. I decided to wait it out until DDR5 got cheaper (I also got Ryzen 5700G for mom) so I dodged so many bullets.
Also, 11900K's AVX-512 is faster than Zen4/5, and a lot of emulators are taking advantage of them now. You can make some really boutique specialty builds with 10900K/11900K Hackintosh/PS3 Emulator
Most people back then suggested the 10th gen with the b560 and screwed people out of their top M.2 slot. All for price to performance. Discord was loaded down with that "Upgrade path" back in 2021.
can you test the 0x12b microcode?
did they actually fix these problems?
I don't disagree with you about people still having issues after this microcode. But I think what you're seeing here is a seperate issue. This microcode is supposed to eliminate voltage requests above like 1.54v during light load spikes. Buildziod did a vid on it
What a waste of time this video was. Especially at the end where he slags off Intel vs TSMC. Completely wild speculation from someone who has done no real physical voltage testing and has no idea what they’re talking about. Buildzoid is who you should watch on this for real insight.
Yea sure buddy. But my CPU is still drawing +1.6v with the new BIOS on defaults. TECH Yes City is right. On paper it should not. But real world tests prove that the voltage is still too high which will still cause degradation.
The VDD Setting is configured by the Motherboard manufacturer, Intel has clear specifications for that (at least in this one setting), maybe because is a Jedec standard. It's been super high allways since 6th gen Intel. I allways tune it down.
Edit: If you want to lower manually you can set it to 1.25v (That should suffice up till 6800MT/s)
It's a great day for some Tech Yes City!
Perhaps the best assessment yet on the Intel issue. Would love to know your thoughts about the difference claimed between the Desktop & Laptop versions of the i9-14900K(desktop) & i-9 14900HX (laptop). Are they so completely different that laptops don't have the problem? Or are laptop users just not as savvy at detecting these issues? Seems like Intel wouldn't re-invent the wheel that much between them, and likely uses the same silicon, correct? Asking because I just bought an Intel laptop & it's thermally throttling despite Extreme Performance fan settings but can't be undervolted. Still have 116 microcode from BIOS IIRC. Definitely not 129. And no, I'm not going to re-paste the chip on a just received, out of the box, laptop from a major brand.
Seeing the wattage and temperatures between updates seems so insane.. Honestly why I've preferred eco-mode and undervolting. Eg, I run my 5950x at 65watt, it runs quiet and cool, and is still a great performer for productivity workloads. Completely agree, bring back the headroom, let the enthusiasts overclock, make OCing a product feature again, and bring back CPUs (and GPUs for that matter) to more sane levels of heat and power consumption by default. Great content!
AMD did just that after reviewers said so and now the same reviewers gave crap reviews of 9700X lmao
On an asrock motherboard where do I go to sync all pcores and ecores. Thanks
Missing sonic theme outro.
is it true that the 13400f revision C0 isnt affected by this flaw?? ive seen in some forums that it is a 12600kf rebranded
DO NOT FOLLOW THE "ADVICE" IN THIS VIDEO!!
- Compared 2022 BIOS to 2024 BIOS, saw voltage increase
- Clueless about the 2022 BIOS undervolting the CPU by 50-100mV out of the box using loadlines
- Thinks Intel is raising voltages using microcode to hide degradation instead of checking said loadlines or the VF table to check what voltage the CPU should have been using
- Gives some awful advice about rolling back to old BIOS, setting static Vcore 1.3V, reduce turbo ratio to 52x, and disabling all the E-cores.
so should i just use the default bios 0x129 microcode update that asrock uploaded?
@@denvernaicker8250 yes keep your bios updated
I haven't had an issue with my i7 13700k and i use the boost overclocking. I guess it is a hit and miss wit hthese processors but I think the first ones made might have been then issue more than the newer ones
I agree. Really bad suggestions. Nobody needs to disable E-cores. Also 5.2 is really really low. You could just configure Per-core multipliers and configure max multipliers to 57 or 58 for up to 6 cores and 56 or 57 for up to 8 cores. Then try settings -100 or -150mv for all VF points for 51x and above. And set power limit 1 and 2 to 253W.
This! Sadly, Tech YES clearly does not understand the issues involved here.
To be fair, when I built my Ryzen 5600 system on an Asus motherboard, auto overclocking was on by default and VDD jumped crazy high on load, up to 1.47V.
When I turned PBO curve optimiser to -15, the VDD jumps to 1.18V max while the processor boosts to 4,65 Ghz (thats 150 Mhz higher than advertised).
But that's ASUS being ASUS. They are known to push insane default voltages on various AMD and Intel systems.
Hey, please take a look at Buildzoid’s videos and compare them to your boards, you two are onto completely different things it seems
How does disabling E-cores help if the problem, as you say in the video, is P-core related? I have an i5 13600K. Should I disable them too? I know they don't help much in gaming, but they speed Windows noticeably.
Spreading misinformation isn't going to help things, mate. 1.424v under light load is NOT "out of spec". The VID tables themselves for almost all 14900K go above 1.424v. Now if you are seeing 1.424v under full load, then you might have an issue. The reason you are seeing increased voltages is because boards before were undervolting by setting the AC_DC loadline lower than Intel's spec. The voltages you are seeing is not what was killing the chips, it was the chips hitting 1.6v+ spikes when transitioning from heavy load to light load.
If you can link me the tables for safe voltages specs that would be appreciated. Other than that I will stick to motherboard guidance, which historically has been accurate.
Also e cores enabled sure. Vdd spikes really high. Though I'm comparing what these new updates do for specifically p cores, and the vid voltages that are assigned.
@@techyescity "Historically accurate" 4096W pl1 and pl2....
Pl1 and 2 limits are wattage limits? Not sure what that has to do with voltage safety limits. You can't bust a cpu with a voltage cap in place and a pl1 limit of 4096. Though you can bust a cpu with a high voltage and low wattage limit in place.
@@techyescity It's just to say that a mobo showing something in red or not has nothing to do with safe settings, 4096W has never been safe but no mobo has ever shown it in red.
Really nice information here. I have a question.
13600k stock(almost) is running at 1.48 Vcore in desktop. In game its like 1.3-1.35 Vcore.
The only thing i did to my cpu is SA Voltage to 1.33 for my ddr4 ram oc.
I tested this for stability like 24/7 and think the ram is really not a problem at all.
But heres my point. Even if i try llc , lite load, voltage offset and any of these combined together im getting game crashes every few minutes. Its crazy. Even only reducing the voltage by 30 mV.
Do you guys think the SA V is maybe causing me this trouble? I would actually set my ram and SA to stock if this could solve this more or less. Maybe im just going to disable the e cores and try everthing off that point and leave it at that.
The thing is im not quite getting pc crashes yet and i really dont want to bother with warranty return, where i wait 2 weeks for my cpu to come back, only because "there's no issue found".
In all fairness, No one but Michael Kelland John Hutchence could front INXS.
Tell me you're old without telling me you're old
@@fajaradi1223 Like fine whine.
just keep walking.
can you share the program how to check VDD CPU
They are afraid that they could lower it and it would be not enough current for degraded CPUS as these would need extra voltage to reach stable point
Yup, that sounds likely why they limited spikes but increased the overall voltage to try and stabilise already degraded CPUs. Slower for hotter, pretty crap.
Hello, I have an i5-14600kf, I put the vcore at 1.2v and individually left the limit on each pcore at 5ghz, the question is, am I keeping everything very low at 5ghz or could I keep it at the factory default at 5.3ghz, remembering that I'm using the latest bios 2503 on a rog strix z790-h, the bios configuration is in the Intel standard without touching other things, I just activated XMP and these configurations already mentioned. (Note: I feel the machine is extremely stable and the cpu is new, used for the first time just now)
this is just one big mess. Some people aren't able to do their jobs since the damn thing keeps crashing on them all the time.
Very well done, Extremely well done infact. My hats goes off to you. Some UA-camrs are tweaking their sh*t for hours and pretend they don't which is a bit disingenious to me. But you stay true to show us what exact settings trigger what behavior. You don't have a biased fanbase which probably helps. Regardless of stupid logo's we are just trying to survive anyways.
Thank you dude! Cheers.
AMD did exactly that with ryzen 9000. They let everyone overclock if they want. But people are mad LOL
I, personally, love what AMD did.
If you read intel statement ; The issue was identified in late 2022, and with the manufacturing improvements and additional screens implemented Intel was able to confirm full removal of impacted processors in our supply chain by early 2024. However, on-shelf inventory may have persisted into early 2024 as a result. I got a new batch of 14700kf and i watched its behavior vddq is locked at 1.4v and v.core 1.3v, it is uv -0.95 and Oc 5.730 ghz p core and 4,4 e cores.
It's been also reported that disabling the Intel baseline profile in the newest microcode bioses removes the new voltage limits and the CPUs can spike up to 1,6 volts on the vcore. This architecture has failed to deliver.
Fully overclockable CPU...if you don't follow intel specs you can fully overclock it, meaning you can use settings that can hurt the CPU.
The whole idea of intel defaults is to make sure that users have a way of knowing that mobo makers use or don't use safe settings.
@@terrylaze6247 True but in this case you can't even be sure, and it's not the mainboard manufacturers that are at fault.
@@terrylaze6247 and intel's defaults are frying cpu's! so yay!
@@dankvader420 It's not the mainboard manufacturers that are at fault for removing the voltage limits when disabling intel baseline?! Whos fault are bios settings then?!
@@Celatra I have not seen a single report of a fried intel CPU (13-14th gen) , maybe you are thinking of ryzen 7000 that would blow up, that you could call frying , for intel they "only" degrade, because they ask for too much vcore which these microcodes are trying to fix.
My MSI board does not seem to have a setting for max VDD CPU voltage
I think I was lucky when I got 13700kf almost 2 years ago. I followed skatterbencher overclocking video I locked cpu voltage to 1.4v. I think this saved my cpu from degrading cause I did try the default voltage but when I saw the vcore around 1.45 in hwmonitor I immediately locked it to 1.4v. I never experienced any blue-screen or anything. This taught me to always control my system rather the microcode runs it wild.
my 13700k was unstable from day 1
@@Adam-jr4lxthat's depressing. Have you managed to rma it? Are you still stuck with it?
@@Eihei Yeah I'm still stuck with my 31.8K R23 13700k after 2 years. Power save mode 18.5K.
@@kramnull8962 isn't 31-32k pretty normal for 13700k? Unless you meant that you have to use power saving mode.
I really hope for you that Intel will improve their rma process, because having to do multiple attempts when a single one usually means that people are stuck without pc for several weeks is insane.
@@Eihei Yeah, but after 2 years I get the same thing. Like I said if I let it beat off 100C.
Vdd is the main voltage fed to the Drain of the MOSFET transistors.
In the case of TTL circuits, which use bipolar transistors instead of MOSFET, the general power supply voltage is called Vcc, where c stands for Collector.
AMD released their 9000 series CPU leaving overclocking to the consumer and *most reviewers are grilling them* so I don't think we are going back to releasing CPU not overclocked from the factory any time soon.
AMD's marketing team didn't help their case with their promotion materials misleading expectations so it is what it is =S
If AMD told consumers ahead of time that it was basically an efficiency upgrade, people would be fine. I think they are great. I was not impressed with the Ryzen 7000X chips. They run too hot and consume too much power.
When it’s an artificial benchmark race selling modern CPUs it’s a race to no where without any major IPC architectural changes .. bode shrinking hasn’t gained a lot of ipc since tsmc 7nm it has only gotten more efficient and more dense … hence the rtx4000 generation. Ryzen 5000 is still very competitive with 7000 and 9000 … they have already reached the limit of the technology at this point they are just refining the process generationally.. with the cpu side the only way to win for day one reviews is to push voltage to hit targets to be at the top of the reviewers cinebench runs haha… it’s kinda stupid but if Intel would of just locked all cores instead of hand the single core boost algorithm to hit the 6.0ghz for single core benchmarks it’s burning up those cores pushing it that hard should of just left it at 5.7 all core and said fuck it if we lose some artificial benchmarks we lose some this gen … but the way incompetence has spread in every industry in the country it wouldn’t surprise me at all if Intel only has the most incompetent of leadership at this point they are so much bigger than AMD who I’m sure are just more efficient and competent at this point where it matters but Intel isn’t being blown out the water really they are tied basically with a lil worse power draw last gen not that big of a deal AMD survived a decade being worse performance per watt haha they should of just buy the bullet and worked on there next generational products instead of refreshing with higher power draw algorithms..
Gamers think AMD exists to serve them.
Their current focus is on datacentre. They make datacentre cores , and sell some to retail.
To use an analogy, think of Zen4 as the standard pentium, and Zen5 as the Pentium MMX.
The p166 and p166mmx were identical in most software.
It wasn't until software was compiled for MMX that it pulled ahead.
AVX512 is basically the same as MMX. It is a SIMD unit. It's wider, and floating point, but the general principle is the same.
So AMD are selling "Ryzen MMX" Datacentre folk love it.
Gamers don't see the point.
In time, they probably will.
@@Austin1990 Same boat here. I will wait for the usual "AMD price drop" as I run some servers at home so I would really appreciate the efficiency gains but not at launch prices =)
@@Austin1990It’s really not much efficient at all.
If I am rendering still images in Blender 3d and Corona for 3ds max, should I disable all e-cores.
Wow, this is horrible advice. For one, youre forgetting and not testing for transient voltage spikes. The previous versions during single core loads were spiking as high as 1.6v, THAT is what causes degradation. The new updates remove the transient spikes. And as for your advice, you dont need to gimp your chip like that. To eliminate the spikes on single core loads, lock your core multipliers to the default boost clock for your chip. This prevents the transient spikes under single core load because it doesnt try to over boost those preferred cores. Jayz2Cents confirmed with one of the SI's that the transient spikes were on single core workloads. And you should be able to get your temps down by using a voltage offset or ac load line to eliminate any thermal throttling.
Imagine going through all that to make sure ur 300$ cpu doesnt burn itself down. And while consuming too much wattage for what a cpu should be pulling.
My i7 14700k ran at 1.387v stock in the bios, I reduced the VRM Core Voltage by 0.1050mv taking it down to 1.285, then reduced the Global SVID Core Voltage by 0.100mv. Cpu shows 1.190v in bios.
Would you guys think this is great? My cpu benchmarks are the same as stock if not 1% better. Intel utility benchmark does not say Yes to Thermal/Voltage limit (can't remember what it was) and current limit, during the benchmarks where as it did when it was on stock, and with just the vrm undervolt it was just the current limiting during the benchmarks.
Problem I believe with the chips is the e cores running on the same rail as the p cores.
Should I update my bios ? I have a new unused i5 13600K i am willing to use, should I use these microcodes and bios updates ? or just undervolt in bios manually, and if yes please let me know how
I wish Intel would just ditch the P core and E core architecture, I know the extra cores look good on paper but it seems to cause so many issues.
I also think this is Intel's Bulldozer moment, as bad as it is, AMD's hands aren't clean historically. Hopefully this is the push they need to get back to making better CPU's
Its the future. AMD will most likely move to it aswell in couple of years
Net burst was Intel's bulldozer, this is an entirely new League.
@@GoonyMclinux
Yes, the cycle of life keeps on turning.
E core has never the issue on this problem it all about overvilting from intel on the P core for example my E core can go up to 4.6ghz without issue and running way lower voltage to boot
Is it worth upgrading the microcode if already undervolted to P1 P2 at 232W, max vcore 1.30V: offset ring -0.1V and vcore -0.08V?
Sounds like the 11th gen cpus are starting to look better.
12th gen is fine and slots into the same motherboards.
I have 10900K, 11900K, 12900K undervolted and zero issues ever.
11900k masterrace. Fastest CPU from Intel that won't randomly kill itself and support AVX512 stock (yes, you can disable E-cores on socket 1700, but that kinda defeats most of its appeal).
12th gens have no issues and AMD 7k series is good from AMD, newer 9k i would avoid till prices drop as it isn't good at those prices vs the 7k series and all reviews show the same thing unless you want to sacrifice all that efficiency for PBO output and that is a lottery to a great extent, but those results can be sen at Jayztwocents YT page currently. I wouldn't be looking backward technology wise but you are free to buy what you like. Cheers!
According to Puget Systems 11th gen actually had failure rates comparable to the 14th gen. 12th gen was the most stable by far.
Hey I got a i9 13900kf have it for over a year now, never updated the bios, got a msi mag z790 tomahawk wifi ddr5, its oc to 5.2ghz all core with undervolt offset - 0.110 and its never pushing more than 1.275v and 200w, max temps in cinebench 86c with score 38k, dont have any stability issues never get blue screen, only time I did get a blue screen was when I pushed the undervolt to high like 0.150v other than that no issues, the question is do I have to do anything when there are clearly no issues and don't really see the point?
I'm sticking with my x99 system 🤗
What 360mm aio is in that 13900k build?
Um hate to break it to you but without an oscilloscope+ counter, the only way to actually see micro code working or not ( Hint its already out and on YT now ). Not going to do that with HWinfo etc apps. I used to do this stuff for a living, Apple, IBM, MS, Game Studios etc. retried hardware eng. Lets set the proper example for folks if you don't mind. While HWinfo is a great utility it just is not designed to do this specific job. We know we been doing this for much longer than a few years =)
Power readings from hwinfo and what's coming out of my watt meter correlate perfectly. I don't think the sensors are faulty?
@@techyescityHW Info has a 20 millisecond interval. You have no idea what is actually being requested by the CPU, you’re seeing a random snapshot.
Was happy to see someone else said it. The sensors in hwinfo is a good guide, but it's not the real picture. I normally like Tech Yes City, but he's downright wrong here. Intel set a hard limit on the spikes to stop them going above 1.55v. The 1.424v under light load is NOT out of spec, it's literally part of the VID table. 1.424v is actually low compared to what I expected him to get mad about.
@@techyescity ua-cam.com/video/SMballFEmhs/v-deo.html educate yourself sir. Not an issue of faulty sensors, they are not capable of the fine increments and speeds of those measurements. You simply can't see nor measure everything properly without one and a counter. By all means don't believe me ask any hardware engineer or Fab builder. Cheers!
@techyescity This is buildzoid using an oscilloscope to measure true voltages ua-cam.com/video/TOvJAHhQKZg/v-deo.htmlsi=vp642zJBKcszb5Ra
I disabled my e cores except for 4 of them. Now when I download files from a 3d website, it takes forever to download them. Is that because I disabled the e-cores? Thanks
at the expense of performance 😢
how does all this relate to stacking more and more cores and trying to power them on die from one rail causing the issue of degradation?
oh and if you had bluescreening already your chip is permanently damaged and you cannot recover it with microcode.
From what I've seen it helps the peak vid table draws but they can't lower it enough because it will eat into performance. That has to happen to fix the degradation and make it stable. Now with this microcode it might help them live longer but it will not fix the problem. The problem is fixed by disabling e-cores and putting the CPU to somewhere between 5.4-5.7ghz. Along with limiting the peak power that the CPU can even request (check Actually Hardcore Overclocking for a video on that).
Long story short, the raptor lake P-cores need to be downclocked to 5.4-5.7 for it to be stable and not shove massive voltage into the core. End of story.
Are laptop processors also experiencing elevated voltage issues? I have an i5-13420H with max TDP of 45W. Should I be concerned about a microcode update?
No, it’s dumber than that. There is a bug in the code that disables voltage limits so you got it going over the safe amount. It’s not a thought out thing, just a rushed oversight
I have the ASRock PG Lightning /D4 board (I think you have the D5 version in this video). I found that setting the CPU Core Current Limit to 310 on my board keeps the speed about the same as prior to the microcode update and keeps the temps down and the VDD CPU never going over 1.336. I also had to disable my RAM OC to keep the VCCSA voltage at stock.. For reference I am using the cheapest 280mm water cooler I could find and it holds at 95c under the Cinebench r23 load... Am I fooling myself, or is this a safe(ish) setting to use...?
Why are you manually changing the Voltage? The whole point is to not do that. Use the default Intel settings. To me this only proofs "somebody" has OCD and just cannot help changing things
Hello @Tech YES City and other viewers .. I have a Gigabyte Z790 UD AX board updated to latest BIOS F11e with 0x129 fix running a borrowed i5 12th gen. I have to upgrade my processor. I wanted to go with i7 14700k but with all these issues going on with instability and processors degrading; do you think it is advisable to go with an older 12th gen CPU like maybe an i9 or i7 of 12th gen and compromise using an older generation instead of opting i7 14th gen to avoid this mess ? what is your opinion in my case. Thanks
He is seeing increased voltages with the microcode. He likely set something incorrectly. The microcode caps VID to 1.55, does not increase voltage.
Both might plausibly be true - you could have lower peak voltage with higher average voltages
On software it says that, I want to see the peaks on a scope because software isn't accurate.
He's seeing increased voltages because the Intel Baseline Profile brings the AC Load Line back up to their spec which causes less vdroop under heavy load. While this gives higher overall voltages, it also helps lessen the extreme voltage spikes. Nothing stopping you from putting a negative offset on your core voltage if your CPU can handle it, but a lot of CPU's, even brand new, would be unstable under the old load line settings.
@@ZoneXVexactly
@@ZoneXV if so, the intel baseline profiles have been around a while, and he conflated the microcode update with a bios profile change. Again, poorly controlled bios configuration invalidates the premise.
You test all conditions equal, not unequal.
Intel has fallen which is so crazy to me...
Stumble at best..This is Intel they are not going anywhere. They attempted to produce their own chips first time instead of all TSMC produced and it didn't turn out great to be honest hence oxidation issues they experienced in fabrication over 13 sku lines. This is why some people are having no issues at all while others suffer greatly.
@@tysoncoffman7562what are you on about? Intel has always fabbed the majority if not all of their CPUs in house, part of their problem is how far they have fallen behind TSMC vis a vis fab processes
@@purplepioneer5644 No they haven't. They in fact make fab themselves in such places as Oregon, Arizona, Ireland, Israel, Design Fab in Folsom Ca, Malaysia as well as other places and contracted work from/thru TSMC in fact they just recently spent 14 billion on manufacturing it new chips. Intel's Mix- Source strategy. Like it or not its a fact. Do go look it up, and educate yourself. Cheers!
Does anyone know which values you have to change on an asus bios, since cpu_vdd voltage is named differently, also for the memory controller but that is a little more eyplanatory
Oh cool the microcode will help the damaged cores heal after the over-volting and reverse the in-factory oxidation defects . Nice !