Testing the intel 0x129 Microcode on the Gigabyte Z790 Aorus Master X with an i9 14900K

[@badhusband1902]

intel validator: ok cpu boots, nice, ship it

[@LorentGuimaraes]

I think they suspected the CPUs were going to degrade all along, but were just surprised by how quickly it happened before warranty and how widespread

[@Infi1337]

@@LorentGuimaraes As an AMD Clown with no logical thinking i would propagate this too.

[@LorentGuimaraes]

@@Infi1337 I have actually never touched an AMD part and have built several computers for myself and friends using exclusively intel processors. It is not far fetched to think that they understood that this architecture was being pushed to its brink. If I as an Intel user all of my life fail to criticize them and suspect of any wrong doing that would make me an intel clown.

[@mattzun6779]

Intel clearly knew that pushing 253 watts through a chip would degrade it faster than pushing less power. If the lawsuits don't move forwards, we may never know if they though the chips would last at least 5 years or 3 years and a day. If I was an Intel validator, I would be heavily focused on the profiles of my big customers (Dell/HP/Lenovo...) Those systems have probably always used PL1=125W and PL2=188W and other sane settings to go with their single stick of DDR5-5600 RAM. Its not unreasonable to expect that there are gobs of test documentation showing that a PC set like that should last a long time. I'd be amazed if Intel expected a large number of failures for OEMs withing three years - they can't afford to annoy them It would also be amazed if Intel didn't realize that there was a risk involved in allowing MB manufacturers to use extreme settings as the bios defaults. They just didn't want the tech sites to show the actual performance of Intel chips configured the way the OEMs do.

[@ArtisChronicles]

​@@Infi1337 So what are you trying to say?

[@johnlong6197]

“End user should not have to fix this”- sums up this whole situation bars😎

[@JustSomeGuy009]

End user didn't have anything to fix, he was chasing ghosts

[@skyvenrazgriz8226]

Well except for the part were every joe has to do a bios update... which would asume they even know about the issue, know what a bios is, care to fix it or let it be fixed and then do so...

[@fergus247]

how much of a task is updating the micro code? i assume its just like any other update process

[@Dragon_Slayer_Ornstein]

Is there any reason this microcode can't be pushed through the operating system via an update?

[@D3nsity_]

Brother if you buy an i9 cpu and have no idea what a bios is you are doing it wrong.

[@PAIN-ot4cj]

Hell yeah the video we were all waiting for, I don't know about you guys but this is better than some movie releases

[@τετέλεστα]

most* movie releases now days..

[@zivzulander]

This could just be the raw oscilloscope capture footage and it would still be better than some movie releases 😁 Buildzoid exposition is a value-add, though.

[@SRSpawn]

😂

[@paul2609]

watch anime instead

[@je5terc0re]

18:07 summary #1 22:03 summary #2

[@fracturedlife1393]

😅

[@afti03]

"I hate this so much...."

[@Infi1337]

LMAO

[@Dhruv-qw7jf]

Testing the 12900K/KS, which Intel claims is unaffected by these issues, with an oscilloscope would be interesting to see.

[@ActuallyHardcoreOverclocking]

I don't have either of those CPUs ATM

[@yarost12]

I reckon those are properly binned, cherry picker CPUs that actually can run 5.5 GHz, unlike 13/14 gen where every CPU is expected to run them

[@Nopanda_Architect]

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 and running at windows 11 pro.. but still have lot problems on there with this SPECS.. any suggestion?

[@merclord]

​@@Nopanda_Architect I'm no expert on configuring motherboard settings, but I have an i9-14000K and I'm on my 3rd CPU. Based on my experience, once you start getting BSOD your cpu is permanently damaged and needs to be replaced. In my case I thankfully purchased a pre-built system with a 2yr warranty so I never had to fight with Intel directly to get a replacement. Having said that, my system was still down for about 10 days with each replacement. On my last replacement I convinced the company to replace the entire system because this issue could damage RAM, motherboard, SSD, and CPU. My recommendation is to RMA your cpu as defective and when you get a replacement DO NOT push your system hard at all. I refuse to run any benchmarks. I just run HWMonitor and keep an eye on the temps and voltage when playing games or running any software that I normally use.

[@toseltreps1101]

I suggest you trash it

[@vladimirrblackrock]

Buildzoid coming through with the real goods!! ty as always dude!!

[@Steve-ho5zj]

I don’t even own Intel and i am watching. Learning stuff.

[@ITSthatMEXICANO]

Same but also as a recent EE graduate, I’ve always wondered how this stuff is measured. It’s also pretty similar to what I’ve done in labs for some of the classes I’ve taken - such as verifying tolerances of components and just making sure our shitty circuits even work.

[@DenMarket-uc1nq]

Seems like we learn the most when stuff goes wrong. In my case, learning is expensive.

[@eliotrulez]

same :)

[@ZoneXV]

Oh, and that bug with the e-cores and CinebenchR15 is a Windows 10 specific issue. I was able to fix it after I did some digging into researching the power plans. You have to use either the registry or "Windows Power Plan Settings Explorer Utility" and set Heterogenous thread scheduling policy to "Prefer performant processors" and Heterogeneous short running thread scheduling policy to "Prefer efficient processors".

[@GeorgeTsiros]

Does that also explain why CB says there are 32 *cores* ?

[@SzcZ]

powercfg /powerthrottling disable /path "\CINEBENCH Windows 64 Bit.exe" That did it for me. You need to run command line with admin rights.

[@Avenge_Computers]

It changes nothing on load but on idle doesn't reach the crazy voltages anymore. They are simply making sure the cpu lives until warranty is over. Still reaching stupid voltage under load.

[@techluvin7691]

Yup…….i agree completely. That’s pretty much the exact conclusion I came to after seeing a 1.6v spike in Jay’s testing on a MSI board I believe. It’s more of a “placebo” than a fix.

[@enrimich64]

That’s what I was thinking too after looking at the data, this cannot be a fix, the voltages are only slightly different

[@ROB-LAMONT]

thank god i just bought my 13700k in june. 5 year warranty im good.

[@fystry]

​@CitarNosis317 To be fair. Puget is top level partners with intel and the company's president is a representative of the Intel Board of Advisors so I wouldn't really fully trust on what they say..

[@DingleBerryschnapps]

​@@fystry Really? I'm going to research this right now. I'm sure that would have been mentioned by other people I follow, which makes me a bit skeptical, but it wouldn't surprise me.

[@aytokpatop]

Feeling lucky because builzoid owns a Gigabyte motherboard

[@rtyzxc]

So previously it was making straight up 1.59V requests, not even transients. After update the spikes cap at 1.54V but is that transient or still actual request? For an actual request it still sounds quite high, though still miles better. I name this generation Intel Volt Lake.

[@AbcdEf-lz6oe]

Nah, that’s boring. Microcode Gate or Volt Gate

[@dethskullcrusher]

Is an actual request, it's shown on video, transients look like a spike, and those voltages look like they're sustained. Remember the fix has to make cpu's not break, it doesn't need to be optimal, in the eyes of Intel it just need to be good enough.

[@jst8922]

@@dethskullcrusher What about ring bus ?

[@imglidinhere]

9:20 It took them a year to figure it out because Raptor Lake went from design to store shelves in 11 months. It took them a year to figure it out because they never did ANY long-term validation due to panicking about Zen 4. WE were their validation testing. >:|

[@celticgibson]

Intel users were the real Beta cucks.

[@Nathan0A]

I used to image a CPU’s QA process like it was a whole team of people intensively pouring over oscilloscope waveforms. In reality it’s a whole team of people intensively pouring over quarterly reports.

[@ActuallyHardcoreOverclocking]

but they should've already had endurance and reliablity data from 12th gen that they can use for 13th gen since they are on the same manufacturing process.

[@CompatibilityMadness]

​@@ActuallyHardcoreOverclocking They have 1.7V as max in specs. Guy doing the testing saw it, went "it's below 1.7V, so it's fine", and the rest is history. Having spec that removed from reality is at fault.

[@stevewatson6839]

@@CompatibilityMadness Teaching Grandma to suck eggs. BZ has highlighted 1.72V in Intel docs on numerous occasions since this began.

[@bazingania2064]

What i have noticed is, that my CP/SP Score in the Gigabyte bios (I have a 14700KF) went down with the latest bios from 93 to 88 and that my CPU requests higher voltages than with the previous bios Version

[@WayStedYou]

if it doesnt spike and cook your cpu but has a bit higher regular voltage it shouldnt be too bad

[@starrims]

Same, mine 13700k from 82 to 76!

[@vbozenko]

@@starrims same for my 13900K

[@3do2]

Same here My CP/SP Score in the Gigabyte Z790 AORUS MASTER bios with a 13900K went down with the latest bios from 88 to 80.

[@robertvansteinshwaga]

Make sure you turn off IA CEP. New bios turns it on by default. When it was off before. Turn it off and you will get your old score back or close to it.

[@adul00]

"Software has bugs", "Microcode has bugs". Oh, please believe me - hardware, unless it's dead simple (think 74xx series chips), has bugs as well. Lots of them. They're just better hidden, often by software workarounds, and "chicken bits" in registers.

[@ActuallyHardcoreOverclocking]

AFAIK intel created microcode so that they could patch hardware bugs.

[@adul00]

@@ActuallyHardcoreOverclocking In strict meaning, the purpose of microcode is to translate complex (as "C" in CISC) instructions to simpler ones, that are manageable by CPU circuitry - as directly executing x86 would be a nightmare. This microcode proper can also be patched (although not in its entirety). The part that manages voltages and such things seems to be rather a form of firmware, internal to the CPU, although the term "microcode" seems to have been borrowed or broadened by Intel here, to cover that as well. David Kaplan (from AMD) has a great presentation on validation of hardware, shown during 32C3 (on YT as well). Edit: I stand corrected: µcodes of today are also responsible for other bug fixes, like related to interrupts or power management.

[@vincentschumann937]

@@ActuallyHardcoreOverclocking microcode breaks down instructions into smaller parts for faster execution and simpler silicon design, they made the microcode updatable around the pentium pro for bugfixes and probably also to make internal testing easier

[@mycosys]

in 74 series chips, theyre just quirks XD

[@mycosys]

@@adul00 That may be what its used for now - but it was not what it was used for on the 486.

[@2xKTfc]

With shutdown /r /t 0 /f /fw you can restart right into UEFI. Put that in a BAT file, put it where the search indexer can find it, and you can restart by typing "uefi" in the start menu and hitting return. No more spamming DEL or F-keys. Even better (and much needed) when you're running fast boot and have exactly 0 seconds to hit the right key. =)

[@andersjjensen]

Thanks!

[@haies09]

Or if you have a stream deck, simply add the windows plugin and set the boot into bios on a button 😊

[@celeriumlerium8266]

You can just hold the delete key, including during fast boot. There is no need for this, if you can open a batch file you probably still have the keyboard and mouse plugged in.

[@2xKTfc]

@@celeriumlerium8266 Not on all mainboards :(

[@DimkaTsv]

​​@@celeriumlerium8266 Wish it was a case everywhere. But no, it isn't. In case of my mobo setting fastboot to "Ultra Fast" basically denies you from entering BIOS. Just because gap where you can press delete is less than 100 ms (aka even by spamming key with 2 fingers you will hardly be able to enter BIOS). So, i will probably just create this batch file somewhere. Would be nice to have in case i need it.

[@wardeagle1015]

I wish we had a BuildZoid "tweak" on an Asus Board for comparison =( Nonetheless love the content

[@DingleBerryschnapps]

Many of the settings translate between gigabyte and Asus boards. Just maybe labeled slightly different. Someone might even chime in here.

[@stevewatson6839]

26:18 I think we'll get a video on just what you want.

[@michelrochette3066]

Looking at my V while watching this after updating with last microcode just to have a better nightmare to finish the night.

[@BaH4o3eH]

May be - I don't know - someone said "Ok look guys, usually our chips lifetime is 20+ years, but we need this voltage so let it be let's say 5 years, it's okay". But it wasn't. It was 1 year unexpectedly.

[@WayStedYou]

more like 2-3 since they have been cooking 13th gen too

[@grizius4123]

That why I never buying intel nor trusting intel ever again. They kept also claiming that i9 14900hx mobile versions weren't affected but that is just wrong. Mine keeps peaking to 1.65v and i get regular 0xc00005 error exceptions on different software apps. Liars

[@IDKYDK-511]

Nice stuff as always :)

[@masoodshah84]

I have an i7 14700K, and it was reaching 1.62V on core voltages, not the package voltage. I fixed it by applying an offset of -0.165V, bringing all cores down to 1.35-1.4V max. After updating the BIOS to the latest version, it improved further, dropping to 1.501V. I then adjusted the offset voltage to -0.100V, and now it stays between 1.3-1.38V, which are much better voltages, and temps are great now.

[@dethskullcrusher]

How you went from 1.38v at minimum load to 1.3v? That's a very big difference, shouldn't be stable using the same CPU. Microcode updates don't increase binning on the cpu's.

[@masoodshah84]

@@dethskullcrusher i am talking about the max voltage it takes while gaming and normal usage i am not doing any benchmarks stuff which is usless for me, just on gaming it was going to 1.62v on core vids, now after bios update and - 0.100V it staying 10c lower and 1.38V max it go in Core VIDS

[@arfanariyanto297]

Same performance and fps ?

[@xupiscodebaleia906]

I undervolted my i7 14700K following the TechSaur guide and the maximum Vcore it reaches is 1.285 V, also the temps goes at 90 °C max (in benchmarks) didn’t lose any performance and also in benchmark it reaches 5,8 GHZ peak. I’m no tech expert but I think this is good

[@arfanariyanto297]

@@xupiscodebaleia906 thx a lot for the answer. Very good news.

[@user-ym5mc]

22:30 Exactly!! I was expecting to fiddle with such settings in order to overclock, not to get it to barely work by default without imploding...

[@serajr]

Thanks for the great info, as always. When changing 'Internal CPU Vcore Offset', there's a note (at the bottom left of the screen) warning that this option is only available if above setting is set to 'Normal', but you leave it as 'Auto'. Just in case! 33:35

[@hughJ]

Paused the video to see how R15 ran and also ran into the weirdly low scores. It also took me a few loops to realize what it was doing. Running a power profile via process lasso that keeps the p-cores from parking seems to fix that, otherwise it's e-cores only. Don't think I've seen another application behave that way. Presumably Win11 manages to negotiate that properly?

[@bogdanc1746]

Intel recommends using Windows 11 with these processors to get the most out of them. That comes down to that hybrid architecture and something Intel calls the Thread Director, which helps the OS schedule posts with greater granularity as to what's going on. Also, Thread Director 2 Has Been Updated For 13th & 14th-Gen. I am still surprised because he's still using Win 10. kzbin.info/www/bejne/iHXXe2ieZbOdrLM

[@Niels_F_Morrow]

Excellent analysis. I think you hit upon the correct answer to this problem. The problem when you have a complete enivronment that swings for the fences, when people are insisting on high frame rates in a game via methods like overclocking, undervolting, etc to squeeze the most out of your CPU, your GPU, your hardware as possible --- this type of storm is created. (o) Microcode to fix over voltage problemsl Preventing future over voltage issues from happening but not knowing the long-term issues in applying a particular fix. (o) Board partners, individual users, generating, creating motherboard profiles that are not within spec. to gain max Q out of the overall hardware, CPU, GPU to up their frame rate in a game or get a higher score on Cinebench for bragging rights. CPU manufactures telling everyone to change MB profile that is within CPU's spec. (o) Users complaining how slow the frame rate is in a game --- how can I increase my frame rate without damaging, compromising my hardware, severely, in milliseconds. Common sense says, dictates this is the kind of environment you would see! What I just stated, what you stated yourself! People are not perfect, you were fustrated about what you thought the fix did but are not sure how this fix will work in the long run. Another problem is not everyone has $200, $500, $600 to throw down, at will, in buying a replacement part that should be working --- supposedly and leads to what I stated above; A vicious, destructive circle based on the idea of getting the most out of the money you already spent to keep up with other people who spent more money on similar hardware to gain advantage over those who could not afford to pay more. Competition is a good thing! Competition leads to change. But when it comes down to what you can afford gets damaged through no direct action on your part. That is where the line should be, must be drawn and why people are upset about this issue. Not everyone knows how to adjust the profile on their MB or adjust one or more things properly where your hardware doesn't get irreversibly damaged in microseconds. Even too much expertise can be a bad thing.

[@Elios0000]

this insane that just running the on auto settings should give at lest stable stock speeds and its running in this kinda stuff.

[@Alonne1]

I have a Z790 Aorus Elite Ax and the first thing i notice with the new microcode on Gigabyte is that there seems to be somenthing weird with the AC LL and LLC previously i could run CPU with a undervolt of Vcore offset -0.088v and Ring offset -0.025v with AC LL of 0.060 mohms (this would be 6 on gigabyte) and LLC on auto with no issues everything was stable. After updating the microcode my previous undervolt settings were no longer stable and even lost 200Mhz Clock speed when the CPU was fully load, I had to set the LLC to medium for the undervolt to be stabe again and then adjust the Vcore and Ring offsets from (Vcore offset -0.088v, Ring offset -0.025v) to (Vcore offset -0.125v, Ring offset -0.050v) and set Vr limit to 1.25V to get the same temps and the same clock speeds that i had with the previous settings on 0x125 microcode

[@Qersoras]

so do you use AC LL 60 and LLC to medium and Vcore Adaptive with Vcore offset -0.125v and IA Vr limit to 1.25V ?

[@Alonne1]

@@Qersoras No, for gigabyte motherboards they use if i'm not mistaken a 1/100 ratio so 0.060 mohms is equivalent to set AC LL to 6 (as an example if i'm not mistaken the lowest profile gigabyte uses for AC LL is power saving and that sets AC LL to 40 that would be 0.40 mohms) besides of that the rest is exactly as you said, but i also offset the CPU Ring to -0.050v, on Intel XTU that would be the processor cache voltage offset under the cache tab

[@Qersoras]

@@Alonne1 yes i have gigabyte, but when i set VR Limit to 1.25 or 1.3 for example , my clocks automatically drop. 5.4 in desktop and 5.1 in games , 36k in cine r23 , but temps are great because in load 1.155 and max 1.232

[@Alonne1]

@@Qersoras Well since you are saying that your clocks drop to 5.4, i will assume that you have an i9 in my case it works becasue i have an 13700k, so if i set the VR Limit to 1.25V i can get the max clock speed in fact i think i can go lower but for other CPU's most likely you would have to set the VR Limit higher to allow the CPU hit those higher clocks

[@Waldherz]

They sure did think that 1.6V is fine, considering how they also say that 1.72V is the peak operating voltage for the 13900K.

[@bigpoppa1234]

Important note for anyone with MSI boards, their new bios seems to set an extremely high automatic Lite Load (at this point in time), causing high stock voltage & heat so you should tune it down and set it to 9 or 10. If you want to do some tuning, most 14700k & 14900ks you can lock the cores to 5.5p and 4.0e with a small negative offset of 0.045, 0.050 or 0.055 and still get high end performance without high voltage & temps. I got a slightly worse score on the new stock bios (with XMP) then improved it with the undervolt while dropping the heat by 20c and voltage down to 1.3v.

[@Pacho18]

That's how i have it rn on 0x123 microcode, and turned back my unstable 14900K to a stable 13900. Then for what you are saying i'm supposing this microcode update doesn't solve the issue yet.

[@jannegrey]

I'd say that is either on MSI or the microcode update was rushed too much (or validation at least). Intel needs to keep MB manufacturers to stay within their "guidelines". It would be great if they went back to "specifications" rather than "guidelines", but that ship has sailed long time ago. Also - most people are afraid to even flash their BIOS. Expecting them to also change stuff in it is a bit too much. Especially since that could and would probably be considered overclocking and push you out of warranty (Intel used to RMA such products regardless, but OEM's not necessarily) - but given how they handled RMA's lately, I'm not sure I'd trust Intel enough to not reject RMA if they know about anything you did in BIOS.

[@bigpoppa1234]

@@Pacho18 No. It does solve the issue, it's just that those new stock settings allow a lot of tuning to run cooler and improve scores.

[@bigpoppa1234]

@@jannegrey I'm fairly confident the spiking voltages are fixed with the bios, it's just that the stock settings MSI applies can make the CPU run hot & push higher voltage loads than need to be.

[@Flodar_Eltih]

Yes but if you set liteload to 9, you have to disable cep. and i have suspection if this microcode update is not a microcode update and just enabling cep defaultly as well as rising liteload mode. I came to watch this video to see how voltage spikes if he turns cep off, but he didnt..

[@WayStedYou]

When the intel cpu designers said "let us cook" that should've been the first warning sign.

[@godwho5365]

🤣

[@Infi1337]

that means they are actually working, not ranting online.

[@goshu7009]

make sens :)

[@alexr8094]

Thanks!

[@ningyang2074]

High voltage can introduce TDDB to destory the CMOS gate insulator. The longer the transistor is exposed to high voltage, the more damage it will suffer. Intel might overestimate the quality of their high-K dielectric or maximal operating voltage.

[@Pacho18]

They knew they reduced their 10nm quality yields as they were worth a lot more to print on early 13th gen. So AFAIK 13th gen is even higher quality yields than 14th gen.

[@GeorgeTsiros]

It doesn't even have to be a voltage that we would call "high". Over time, even moderate voltage will create a conductive path inside an insulator.

[@ningyang2074]

@@GeorgeTsiros Intel might gives a gate voltage just slightly higher than a safe value they used before. Accroding to literature, the charge energy is the squre of applied voltage, and the relationship between TDDB time window and energy can be in orders of magnitude.... A little bit exceeding voltage will reduce lifetime of gate insulator dramatically.

[@GeorgeTsiros]

@@ningyang2074 with how small transistors have become and the weird shapes they now have, the engineering community has very little knowledge about how they fail or behave. No longer are they two rectangular doped regions separated by a line of poly.

[@Mike500]

Nice one as usual. 21:33 At points like these, the audio resembles "2001: A Space Odyssey", when the guy retreives the supposedly defective unit from the antenna module and all you hear is the breathing noise.

[@nonetrix3066]

Watching you go in settings to try to fix things is crazy, I don't know what any of it means. I totally agree there is no way a normal user or hell even a technical user such as myself which uses Linux and has some programming experience should have to know that. I am not a electrical engineer overclocking expert, I would have assumed my CPU was defective or the micro code update completely nukes performance

[@pm110978]

Thanks mate for the video 👍

[@SamOnKBD]

lol your explanation of how VID requests work at the start is giving me flashbacks to wireless Qi chargers. I made a receiver as part of a college course, so you could put our "box" on a Qi charger and it would actually receive power. Basically, you send control error packets back to the transmitter of how much power you want vs what you are getting. And then the transmitter uses a PID controller algorithm to regulate the power you are getting. Both transmitter and receiver have clearly defined specs they have to follow, but I was not really getting what I wanted, so I had to fudge the numbers in the control error packets I was sending back to the transmitter lol. It seems that the motherboard/CPU is doing a similar thing here, where both of them are adversarially working around the spec to try to get what they really want out of each other.

[@GetJesse]

At some point they probably fired the guy that was making sure all this was working properly.

[@MCentral8086]

Thank you for the video buildzoid. Last week I followed your suggestion and enabled the IA VR Voltage limit control to 1450, and undervolted by -0.1V, which gave a cinebench R23 score of near 40,000 and I am happy with that. Question - Given Intel's quality control has proven to not be stellar of late, would you recommend re-enabling the IA VR Voltage control (that you discussed in your previous video) after the bios update to something a bit lower? Intel's new 1.55V cap still feels too high to me.

[@jonathanmartins7744]

8:25 I and a lot of people would love to watch an 1 hour explanation on why does the CPU requests such high voltages.

[@karlos1060]

This is hte reason i did not switch yet to the new microcode. Last time switching to the 0x125 code gave me higher voltages. So with a lot of tweaking it runs perfect again with undervolting and good and fast scores. So i even might not go to the new code at all unless they really have something solid. My I7-14700KF runs CB R23 on max 202 watt and 33.5 to 34K points every time with 85c max. On that task it is averaging at 1.1 volt. Single R23 2150 and about 40 watt max with low temps. My Vcore does not exceed 1.32volt anymore after long hours of gaming and recording gaming at the same time.

[@arypramudito2963]

[UPDATE] Ahh i saw voltage spiked after cinebench r23 finish rendering 27:00. So the all cores like doesn't have anything to do. the voltage was too much where the cores trying to breathe and low down temp. The voltage in hwinfo only 1s sampling, while the osciloscope is 1ms sampling. The Hwinfo only captured 1.49mv , while osciloscope captured 1.53mv 17:23, so expect higher voltage from hwinfo. My asrock z790 gonna update tommorow. Now we had 40000 r23 i9 14900k with intel performance setting, all p-cores boosted not auto, undervolt adaptive, -50mv in 0x125 bios pl1 253w pl2 253w auto default loadline ~ 240mm cooler setting on bios. max 1.475v in hwinfo. -100 mv undervoltage will get instable to cinebench

[@kpm4620]

I really didn’t expect you to look like a Motley Crue fan. I was thinking more like zlaner. Thank for not looking like zlaner

[@WhiteWolfTechGaming]

You go further in depth and explain a lot more than others on here about this new 0x129 - and I'm here for it. But I do have a question about CEP. In your other recent video for this Gigabyte board, the settings for the CEP in BIOS were default to Auto, which stated that Auto was set to “Disabled.” Is that also the same issue with this new microcode? Would it behoove the user to just change that setting from Auto to Enable?

[@sp00n]

On my MSI board, Auto was enabling CEP.

[@martinmarinov1334]

Thanks, this is really helpful, will be installing the new BIOS soon and testing your settings. Hopefully this nightmare ends.

[@c.sec73]

It surprises me that my wife doesn't want to watch this with me.. literally best release for Sunday. Thanks for your testing and research.

[@FyrestarCom]

Thank you so much for these videos, i watched the 2 hour long one 3 times since all you post around this is way more informative than literally any other source online, especially since it's not theory but practical demonstration and testing and tweaking, plus your super friendly voice. Luckily for me you also reviewd Gigabyte since all MB vendors are different. I'm going with the "lower" Performance profile and the IA Voltage Limit of 1.4 you suggested since i also underlocked a bit anyway. What main settings besides downclocking would be the best for longlevity? If i just go super gentle for a couple years i could technially just even out any degredation that might happened? (used PL Limits since beginning, didnt game anything ever on it, owning since 9 months in daily use)

[@n1kobg]

From the voltage ranges you can see Intel tried to design & create 13/14 gen to be more robust than the previous generations. And it is more robust indeed, just not as though as they thought.

[@fracturedlife1393]

18 mins in. Thats a HEAVY sigh.

[@ronaldgutman5980]

Nice Overview Sir, hopefully all those with Toaster Oven Cooked CPU's will get replacements from Intel. Always enjoy watching - keep up the Good Work!

[@NINEWALKING]

I have MSI mainboard, and they have beta bios with 0x129 from yesterday or the day before yesterday. I have said that I am going to try it as the last ditch effort. Now I just remember how many hours I have waisted juggling load line settings and power settings and undervolting and power limitations and current limitations just to get my system performing as it should and being stable. BTW, it never ever was 100 percent stable at long periods. I was one month on vacation, and it was good times not using my main system. Now I see the same random crashes as before. Once in a day or two, the application just crashes. So, I guess my CPU has degraded. But it might have been just to high voltage spikes as well. My cooling is stupidly oversized, and CPU was never seeing high temperature. So there is some hope. I just do not want beta bios anymore. I do not want uncertainty. I do not mind fine tuning it. But it should just work. I do not want to be their tester. I want top performance and 100 stability as promised and paid for. These issues wouldn't be tolerable even on 200 bucks CPU.

[@Pacho18]

I still have 0x123 microcode and just by disabling all thermal and voltage optimizations and capping freqs to 5.5ghz, turned out my unstable 14900K back to a stable 13900 like in perf. So seeing other MSI owners comments on 0x129 microcode saying they're still experiencing issues, and that stock settings are insane, i suppose those beta BIOS didn't solved any issue yet.

[@NINEWALKING]

@Pacho18 I have tried that, and it still didn't work for me. I am about to try the beta now.

[@Pacho18]

@@NINEWALKING i'm at 0.85v on vcore idle

[@NINEWALKING]

@Pacho18 Currently, I am out of the house. I have no clue, but I think idle voltage is stupid high. I really dread that I have upgraded from 12900K to 14900K.

[@Stu_8]

Comparing a Z790 UD AX i7-13700k under the F5 and F11e bios has been interesing Cinebench24 drops from 1646 to 1503 (the value on the Cinebench site for a 13700k is 1608) The old Gigabyte optimisation ran the pcores at 5300 except for temperature throttling, but the new Intel optimisation sees them typically at 4800-4900 due to lack of voltage. Hence the score. The old setup gave the odd VID in the 1.40s in HWMonitor (I was using IA VR voltage 1390 to minimise that but some spikes were 1400-1407). The new setup gave 1.44-1.48s which I found alarming, but still wasn't sending enough voltage to the pcores. I tided the big VIDs up with IA VR voltage 1440 which keeps the max VIDs to the supposed processor voltage max (1.32+5%=1.386) Will be interesting to see if bios updates or tweaks wil get the pcores working harder

[@Hogdriva]

I’m going with “they thought 1.6v is fine” because this is the only way they could compete with AMD with Alder Lake++

[@billymania11]

I remember thinking when OC became a thing that at some point, the whole concept would implode when the node process hit a critical ceiling. It seems we are there. I wonder what happens now?

[@jakubjanicki9148]

"What did they do to the ac loadline this time" 😆

[@yinwachan6329]

Hi, any ideas that my 14900k only scored 38k? I followed your AC loadline setting to 55, CPU Vcore Loadline Calibration is High, added Voltage limiter to 1400 and Vcore offeset -0.13. Im using same motherboard too:/ The only differences are i have XMP on and running on Windows 11

[@Ray88G]

Intel should see this .

[@betag24cn]

intel? they made this crap, they saw this and will try to avoid lawsuits by blamimg mobo manufacturers or any other crap they come up woth

[@grizius4123]

​@@betag24cnthey also denied that mobile cpu are affected of this fking liars

[@andersjjensen]

New video from Buildzoid about 0x129.... and it's 39 minutes. Me: "Oh... Not good!" Buildzoid: "Why would it be easy? I hate this so much!" *Furiously starts debugging* Buildzoid: "Oh... Windows is just being Windows"

[@KoseChris]

the code brings back the turbo boost bug that was fixed in the bios before this one... the 2 best cores don't boost higher then the other cores anymore

[@fracturedlife1393]

"how to update like codemasters"

[@ActuallyHardcoreOverclocking]

I still get the 6GHz boost on the 2 best cores.

[@GeorgeTsiros]

@@ActuallyHardcoreOverclocking Ah, that's because your CPU has the _best_ best cores. Golden sample, etcetc

[@andersjjensen]

@@GeorgeTsiros Except BZ has stated many times that his particular CPU is crap that only just met the binning requirements :P

[@Infi1337]

not on my 13700KF, p-core 4* and 5* boost a bit higher like usual.

[@rayw8252]

Your R15 performance is a Windows 10 and Intel P+E bug. I saw it on the i9-12900K as well when I initially thought "Windows 10 is fine". But after installing Windows 11, the problem went away.

[@starrims]

Can I apply all these setting you showed at the end of the video to an 13700k? 55-55 AC/DC loadline?

[@edplat2367]

Every day I regret getting an evga board. No bios updates because they stoppes employing their bios team. It really sucks.

[@heyguyslolGAMING]

MSI & Asrock imo are currently among the few mobo comps that you can trust when it comes to features, updates, and warranty. EVGA use to be legit but unfortunately fell off w/ the start of them abandoning GPU's. I think we all know to stay away from Asus & Gigabyte at this point.

[@GhostPsi69]

Dont regret, just undervolt your CPU and it should be fine

[@Supernova094]

Do manual voltage and power control and you should be good. But yeah been using MSI motherboards for over a decade now and no regrets.

[@Dhruv-qw7jf]

Asrock is actually just a subsidiary of Asus. ​@@heyguyslolGAMING

[@TwiztedJugallo]

@@heyguyslolGAMING Asrock and MSI are the ONLY legit mobo makers right now. Notice how everyone who's crying about "degradation" is mostly on Asus and sometimes Gigabyte?

[@khensationalReviews]

My 14900K and 14700K with MSI Microcode still is requesting higher vids and using high v core still. Id still sync all my cores to spec and get rid of the single core boosting and do an undervolt. I do agree that its stupid that the processor out of the box settings should just work properly and safely.

[@Pacho18]

Then i won't upgrade. I already got my 14900K back to stable with my settings, so i suppose if you're telling that, that it isn't solved.

[@LooneyTo0n]

Buildzoid, I have a question regarding the undervolting. I'm currently running a 13700K on a MSI Z790 MB, with AC=DC=1.1 mOhms, IA CEP enabled and LLC on Auto (which is mode 8 on MSI, highest VDroop), with an offset of -140mV which applies to both Core and Cache. This is stable and I'm getting very good results, with max VCore not exceeding 1.34V, at least according to HWInfo. VCore under R23 load is around 1.18V and the score is great. I noticed you dropped AC and DC to 0.55 mohms, calibrated the LLC accordingly, and then applied the voltage offset. As far as I understood your explanation, you have to keep AC=DC with a corresponding LLC so that IA CEP doesn't interefere, so my current config matches this guidance. However, what benefit would I potentially have by reducing AC and DC manually and then finding the correct LLC mode, compared to just leaving them to 1.1 mOhms as I have it right now?

[@sp00n]

LLC Auto was mode 8 for you? IIRC setting it to Auto gave about the same Vdroop as mode 3 for my MSI Z790 board, so very far from mode 8 with highest Vdroop.

[@LooneyTo0n]

@@sp00n I think it depends on the AC/DC load line settings, at least it makes sense that they're correlated. However, as I have the AC/DC set to 1.1 oHms, LLC = Auto gives me basically the same or almost the same results as Mode 8. I read somewhere though, that after an inquiry to MSI, they confirmed Auto is calibrated for Intel's recommended DC load line of 1.1 mOhms for Raptor Lake, how true this is I have no idea.

[@haies09]

Im on the Z690 Edge WiFi paired with 13700K and basically the same settings, but I never changed the AC/DC, it’s on auto which is mode 9. PLs 125/188 ICCMAX 307A AC/DC 40/90 Negative Offset -0.150 IA CEP Disabled In R23 I’m getting 30200 pts and barely breaking 70C. I’m sure it’s not stable if I run other benchmarks but all I’m doing on my PC is Photoshop/Premiere and playing games and it’s been running great so far, knock on wood 😊.

[@sp00n]

@@LooneyTo0n Hm, I'm currently not at home until wednesday, so can't look at my notes, but I do remember that for a fixed Vcore of 1.3v mode 3 equaled Auto LLC. However, the AC/DC loadlines were also automatically set to 0.01 mOhm when using "Intel Defaults" when you set set a manual Vcore, and I can't remember out of the top of my head how the Vdroop values where when I used Auto Vcore instead (AC/DC was then automatically set to 1.1 mOhm though, that I do remember).

[@sp00n]

@@LooneyTo0n Ok, I'm back home now and was able to check my notes. Indeed, for an Auto / Offset Vcore setting, the Auto LLC setting results in pretty much the same voltage under full load as when manually selecting level 8. However once you enter a fixed Vcore voltage, this changes to level 3 instead. Kind of weird, but maybe it's indeed depending on the AC/DC loadline, which as mentioned is being changed from 1.10 mOhm to 0.01 mOhm when you switch from Auto/Offset Vcore to fixed Vcore.

[@j340_official]

Is the bug in intel’s raptor lake microcode also present in predecessor CPUs? Like how long has this been a problem for the cpu voltage requests ? Its rearing its ugly head now given raptor lake was factory-pushed to the brink of stability.

[@ZoneXV]

Is there really any issue just disabling CEP if you have a voltage cap? I'm not even 100% sure what CEP does other than screw with people's stable undervolts. On my ASUS board, just leaving the LLC to the default of 3 with no undervolt triggers CEP. I have to raise the LLC to 5 to not run into CEP.

[@he1go2lik3it]

In the Task Manager you can disable the "efficiency mode" for apps like Cr15 if Windows is managing it incorrectly or undesirably. Process Lasso is my favorite for this.

[@chrcoluk]

Your CPU seems to be working much better now, if I remember right? at the start of all this stuff it wasnt able to do r15 at stock? and now you running it undervolted fine.

[@Spinelli__]

I'm wondering, if you're manually tuning voltages, power limits and power options (max wattage, max amperage. CEP, etc.), temp limits, disabling 2-core boost (which pumps CPUs full of voltage regardless of power/temp limits, etc.) then what does a CPU microcode change that couldn't be changed/tuned/edited in the BIOS with the older microcode? The reason I'm asking is because, even after lots of research, I still can't find the difference between BIOS changes/settings and a CPU microcode change. I understand that the CPU microcode is sort of like a firmware for the CPU but if all those changes can be made/tuned/edited in the BIOS - even with older BIOS & CPU microcode versions - then what does the CPU microcode update actually do differently?

[@tmsphere]

You can't edit vid requests there's an algorithm that calculates vid requests that are set during chip manufacturing/testing. A microcode modifies the algorithm to fix manufacturing bugs post release.

[@Spinelli__]

@@tmsphere Ya but you can edit voltages regardless of VID request and even offset voltages based on VID requests. If VID requests 1.35v for the 5.7 GHz all-core turbo (just a made up example) but I know I'm stable at, say, 1.32v, I can just set an offset of -0.03v. I'm obviously simplifying it due to vdroop, LLC, different types of loads, etc. So how does the microcode update change any of this?

[@AySz88]

This may or may not be what you wanted but: Consider the case you kinda identified by omission, where default or auto settings end up doing something wrong. But there are also automatic settings intended to adjust in response to other settings, and things under the hood that are never really meant to get shown to the user. Like every user interface, settings are abstractions, with different meanings and different intent even when they overlap. Overriding every single hardware-level option would be annoying, error-prone, unpleasant... So things get abstracted together, or sometimes split apart. In most user interfaces, a lot of nuance never gets exposed (an "abstraction barrier"). Think like "XMP timings" for memory, or "fan curves" where you edit a graph and not a table. Notice how buildzoid speaks about how certain settings change a voltage-frequency table under the hood, but he doesn't directly tweak individual numbers in it. The BIOS/firmware/microcode is what translates abstract settings into messy practice. If the implementation from user UI setting to hardware goes wrong, that's where the update comes in. (And of course in this case, even some auto settings were bad.)

[@AySz88]

​@@Spinelli__Whelp I missed the new replies. So at that level of detail... If I understand what's happened right, it's not just about the values in the table, but how the chip determines which table entry to use. Some kind of predictive algorithm is guessing ahead that it should request a voltage appropriate for a particular combo of (frequency, temperature, load, power target, ...). But since it's predictive, that (for example) load might not come to pass. The value it grabs ends up being unsafely high when the load is suddenly no longer there, resulting in no actual vdroop. Meanwhile, with low current draw, the actual voltage delivered shoots up, because of the request that was attempting to compensate for vdroop that doesn't exist. (It might be more complicated than this, but it's the simplest way to manifest this error I can think of.)

[@AySz88]

I forgot to mention, there are ways to mitigate this using the available settings, but nothing that directly targets the issue well enough that it doesn't also unnecessarily/unacceptably cut performance. (At least, that's what people are interpreting based on the fact that BIOS updates with new defaults haven't solved the issue.)

[@Dhruv-qw7jf]

32:47 Can you make a guide on how to undervolt these CPUs from within the Asus BIOS as well? My 13600K is fairly regularly hitting around 1.4V and it really shouldn't, considering that it was working fine at around 1.25V before all of the recent microcode BIOS updates that are supposed to fix the instability.

[@AaronWOfficial]

Shouldn't be hitting 1.4, mine is hitting a max of around 1.35 (I have same cpu + asus board). I fear your cpu may already be damaged

[@Dhruv-qw7jf]

@@AaronWOfficial bruh that sucks. Today I saw it take up to 1.480V on the motherboard's Vcore sensor in HWinfo after a Cinebench R23 run. Ran under the Intel Default Settings. I have an Asus TUF Gaming B760M-PLUS. Since it was caught by HWinfo which does not update as frequently as an oscilloscope, I wonder if the actual spike might have been even higher than 1.48V.

[@AaronWOfficial]

@@Dhruv-qw7jf HW info + other apps don't display sensors correctly while idle. My cpu is undervolted to 1.18v, I can even see that it is 1.18v while idle in the main bios screen, but HW info + other apps such as cpu-z\core temp etc shows it is 1.25 - 1.35 while idle. Soon as I put any kind of load on the cpu, the voltages drop to 1.18v in HW info. I recommend you go to ai tweaker and set vrm to offset mode, set offset sign to - and lower your offset down to around 1.18-1.19 v. Note that the offset is subtracting from your VRM voltage, and applying it to your vcore. For example my VRM was 1.25, I did a 0.0700 reduction from that and it got me to 1.18v on vcore. It should be very stable, should barely effect your cine scores, and will lower your temps significantly. Also consider lowering Long Package Duration to 125W, and ensure the short is on 181W. This significantly reduced my temps to the point I am never peaking above 75 degrees, and my performance is more or less the same as stock. I actually have plenty of headroom to overclock now if I want to.

[@Dhruv-qw7jf]

​@@AaronWOfficial Do you mean the Actual VRM Core value should be set to Offset mode? The Global SVID and the rest of the values can't be set to offset mode on my trash-tier Asus TUF Gaming B760M-PLUS. I even tried undervolting from XTU and it said Undevolt Protection is enabled and that I have to disable it in order to undervolt. On my motherboard, only the Actual VRM Core value has the option to be set as Offset mode, the rest of them values are only either Auto or Manual mode.

[@AaronWOfficial]

@@Dhruv-qw7jf Yes actual vrm core when set to offset will give you the option to set +\- offset sign. You set to - sign, and then enter a decimal to reduce voltage by. If you read what the actual vrm core was before you change to offset (it should have the value written right next to it), just keep subtracting decimals from that until you get to 1.19 or 1.18v and try each. Then underneath offset sign you will see option to change core voltage. The decimal points you subtract from the actual vrm should be put here, save & reboot and take another look at bios main screen check if the voltage is right. Then run a 30m stability test in cine to check for crashes while also monitoring temps see if you are happy with them.

[@sp00n]

How can you match your AC/DC loadlines to your VRM loadlines? What's the procedure?

[@DeSinc]

the e-cores in background thing is very confusing and is disabled only by 1. switching to 'Balanced' power plan in the legacy windows control panel power settings, then: 2. "Best performance" in the new windows 11 power settings menu.

[@warrenslater3709]

Dude you need to say "like" more please

[@afti03]

like

[@4epa1012]

Loool

[@Rotwold]

"uhm..." "the thing is..." "this isn't the most..."

[@MhillPlays]

After my experimenting i discovered something weird, manually tuning AC/DC_LL to 0.55mohms caused the vcore to do 1.68 to 1.7v during Boot up. Well that's what my display was saying on the Z790 Formula (hoping it was an error and not damaged my CPU). I quickly reset and won't do that again. I was trying to gain some performance back, but i cannot surpass 35k-36k on R23. Temps are approx 75c under 100% load and i seem to clock down to 5.2Ghz sometimes as low as 5.1 and 50. PL1/PL2 are both 253 and 400a. Disabling CEP did virtually nothing on my board. So yeah, don't manually tune AC/DC_LL on Asus, unless CEP being enabled caused some bizarre anomaly on this Beta Bios.

[@elita2cents]

About this microcode thing, I got my 13700K last August and I was watching HWInfo and noticed some insane voltages being pulled at stock clocks, so I wrote Gigabyte all about it and they said, I should make a screenshot and they would get back to me. That was last October. Unlucky for me, at the time, the screenshot was shot, it didn't show the 1.588 V, that I noticed days earlier on my Aorus Pro X Wifi7, which would be the Master X, without the debug-LED. Gigabyte's reply was, that they didn't notice anything wrong with what my screenshot showed and only one year later, it turns out, I was right. But, what do I know. I am only building computer since 1996. So, I are just a stoopitt customer. Speaking of CinebenchR15 running on the E-core, I was playing Diablo 3 on my replacement 13700K and to be on the save side, I demoted the Raptor Lake to be an Alder Lake CPU with my E-Core turned off and an upper voltage limit to 1.200 Volts and a max. P-core clock of 4.8GHz, just like my 12700. So Diablo 3 was running fine and smoothly but yesterday, it felt really sluggish somehow. That went on for hours and all of a sudden it stopped be sluggish and running smoothly again. That is with the current bios F4e with microcode 0x125 still.

[@Lordssr]

I also noticed that the computer gets slow at times. I also have an I7 13700k and I solved it when I asked Intel for help and they recommended these two options: Solution A): In BIOS, select "ADVANCED MODE", in the Tweaker tab, locate the CPU Vcore and select "Normal" option, select "Dynamic Vcore(DVID)" option, change it from "Auto" to "+0.005V". Increase the DVID by +0.005 and reboot OS until the game crash disappears and the system is running stable. Solution B): In BIOS, select "Tweaker," select "Advanced Voltage Settings," select "CPU/VRAM Settings," adjust "CPU Vcore Loadline Calibration," and recommend starting from "Low" to "Medium" until the system is stable. I'm not an expert, but maybe these solutions will help you or you can explain to me how the problem was solved since you have more experience. (In my case, applying both solutions at the same time made the system stable again and games like The First Descendant became stable again.) I don't dare to apply the latest BIOS released by Gigabyte yet, since I read that several people had problems such as having to access Windows from Safe Mode, so I will stay with this configuration for the moment.

[@elita2cents]

@@Lordssr Cheers man! I left those voltage settings at "auto". I'll try those instead. Thank you!!

[@Pandemonium088]

Im banking on that the IA VR limit on Asus board is effevtive at limiting any and all spikes. My 13900K has none of the stability issue, I would still take the steps to prevent damage, while waiting for Asus to release the final version for my board.

[@timothygibney159]

My hero z790 despite a vr limit of 1700 still degraded my 13900k with a .6 volt undervolt

[@Spinelli__]

I have unlimited power (current, wattage) limits, most or all power limit options (CEP, etc.) disabled, and temp limit set to 105 degrees C. However, I also have the insane, useless 2-core boost which pumps like 1.45-1.60+ volts into CPUs (regardless of power limits) disabled, along with e-cores disabled, along with manually tuned voltage/voltage offset and turbo cores synced (it still fully downclocks and downvolts though). Using those "rules", I have had a 12900KS at 5.33 GHz all-core (5.0 cache) for 2 years, a 14900K at 5.7 GHZ all-core (5.1 cache) for almost 10 months, and a 14900KS at 5.9 GHz all-core (5.2 cache) for almost 5 months...all with no degradation whatsoever.

[@Pandemonium088]

@@timothygibney159 0 A .6 undervolt is massive, you probably did not degrade but just a lack of stable voltage. Imagine if your CPU needed 1.390 to be stable but its only getting 0.79v

[@96kylar]

@@timothygibney159 that isn’t degradation, it just that is a insane undervolt.

[@DingleBerryschnapps]

​@@timothygibney159 Are you sure?

[@jeffamunoz]

At this moment in time with this latest microcode update, the 0x129 - would it be safe to put a sealed i9-14900k in a motherboard?

[@gvido2923]

Yes

[@WayStedYou]

much less likely to be cooked running on 129 the entire tiem than one that has been used but If I were in that position I would be returning it for a 7800X3D or something

[@whoshotya117]

So just lock the cores, undervolt and underwatt?

[@filo2848]

Mine was really degredated, damaged it couldn't run anything without manually tuning the ac/dc and LLC, now i updated to the 129, so i hope i can finally rma it to get a fresh one that won't be killing itself, but the current cpu on 307a runs just fine never surpassing 1.5v but on 400a it goes crazy 1.558v on de desktop and 1.617v for R23 single core on 5.7ghz! Not even the 6ghz boost, and some workloads are still unstable at 400a like unreal engine 5 shader compilation and decompression

[@chromerims]

2:56 -- "The reason this happens is that Intel is trying to predict V-droop before it happens. The problem with this whole idea of predicting V-droop before it happens is: If you predict V-droop and then it doesn't happen, you get a bunch of extra voltage, which is what we're seeing right here." Thank you 👍

[@chromerims]

30:12 -- "We haven't gone over 1.55 volts. It looks like Intel's new [0x129] microcode does what it sets out to do. 35:32 -- "If you're undervolting, you might cause instability by just setting the undervolt too low. (We're back up at 40,000 points [on Cinebench R23].) Right now, CEP is turned on. You don't need to turn CEP off in order to undervolt unless you're going to undervolt by intentionally setting the AC load line below the load line of the voltage regulator on the motherboard, which you can do. But I don't see the point in doing that because if you're using a flat offset like this, it offsets all voltage points the same right. Everything, like the 6 GHz voltage point, gets lowered by 100 millivolts; every single frequency gets lowered by 100 millivolts. 36:29 -- "Whereas if you undervolt by setting AC load line lower than VRM load line, you end up with an undervolt only at very high current output. You won't undervolt your high frequencies while undervolting your low frequencies, which you can do. But I don't know why you'd bother with that, especially since I don't know if turning off CEP is . . . how bad an idea that is. 37:11 -- "Intel does recommend that you keep [CEP] turned on. After all of this, I would follow that advice from Intel. Don't turn CEP off, and you can still undervolt using regular VID offsets. You can't undervolt by intentionally setting the AC load line below VRM load line. 37:40 -- "Now our voltages are way lower because we also have that undervolt across the entire VID table. As you can see it maxed out at 1.4 [volts]." Excellent job 👍

[@krasi_rkkrtabakov4953]

AWESOME VIDEO

[@augustomartins9516]

Buildzoid doing his thing, giving specs and conclusions to techtubers

[@geph13]

when you said these cpu's request 1.6v i literately yelled "GODD DAMM"

[@TigonIII]

I can't believe how engaging these almost 40 min of watching an oscilloscope jump up and down. 😆

[@deepSilent0]

I’m running your tweaks from your previous video and my system is stable and I am quite happy with the performance. Is there any benefit in applying this new bios with the microcode update?? I’m kinda paranoid since it’s been a nightmare in the past…

[@andersjjensen]

In case you need to RMA it anyway it's a good argument to say "Yes, I updated my BIOS on the day the 0x129 microcode update was out for my board".... And you can always roll back if you can't get it stable.

[@chik0240]

Late to chime in, I am using the Z690 UD AX DDR5 with the 14900k, recently I've fine tuned my undervoltage with point selection, and ended up the highest performance setting I could achieve while stable for R15, R23 and 2024, 30min OCCT CPU test was -10mv from 44 to 56, -50mv from 58x and 60x, with auto LLC at 0.4/0.9 with the R15 getting 6070 CP, R23 single ~39786 and 2024 at 2302, any lower voltage would run R23 and 2024, anything more aggressive even with 1.1/1.1 LLC input and manual undervolt will either crash R15 consecutive 5 runs or 2024 crash, or have a single WHEA error in 30min AVX2 OCCT, and curiously, R23 get highest score from the more modest undervolting, the VID cap is around 1.38v and average 1.1 at 5.4Ghz in cinebench and 5.7 all core at gaming loads, I am suspecting that undervolting or LLC being too aggressive would make the CPU hitting random, but not fatal error so draging down the performance and barely not crashing

[@od1sseas663]

How did you get the p-cores/e-cores to work well in Windows 10? On Windows 10, my 14600K gets half the score vs Windows 11

[@Nathan0A]

I would love to see you get your hands on a known degraded chip and show how it responds to undervolting. Also reading something about vMin getting automatically raised on degraded chips, is there a way to check what vMin is being used on a specific chip?

[@Dhruv-qw7jf]

22:30 aren't you supposed to run the "load optimized defaults" right after a BIOS update though?

[@jedenzet]

When you flash bios it automatically sets everything to default

[@Dhruv-qw7jf]

@@jedenzet oh I didn't know that, it might just be some very old advice that isn't relevant today then.

[@haginn]

Thank you for this video. I have a 13900k on a Z790-E Asus board, MCE off. No problems so far.. i have a BIOS from way back February 2023 as i only updated it once for XMP support then never again. AC/DC LL is .5/1.1 by default. I don't know if those are going to change with the update. Do you happen to know the VRM LL for this board?

[@jannegrey]

I think AC LL and DC LL being different is a no-no. Especially when a lot of protections aren't turned off as default. But then again it's only what I heard from many videos, I don't have those chips by myself.

[@Slava4243]

Please correct me if I am wrong here. The fixed (static) voltages make the system ignore VID requests, right? Given the potential risks of degradation and system stability, why should a user use AUTO settings or adaptive settings? I think that it would be more reasonable to set up fixed/static voltages below 1.4v with the highest LLC settings to prevent voltage drop under the workload. The only downside is higher power consumption when idle, but who cares? I do not have an oscilloscope to verify if there are any voltage peaks if a user sets up static voltages. That is probably a question for @Buildzoid. What do you think folks?

[@Pacho18]

How we set those static voltages?

[@Slava4243]

@@Pacho18 For gigabyte MB -> Vcore Voltage Mode = Fixed Vcore (33:11 on this video), and do NOT use AUTO anywhere related to voltages if possible

[@nlgatewood]

I'm wondering if this "bug" was initially intentional..and they didn't realize the damage it would cause..or looked the other way to try and squeeze out more performance. When I looked on my spec docs from Intel for my 13600k a while back, they list the max voltage at 1.72V..which I thought was insane

[@VladMX]

First of all, thank you for the very informative and helpful videos on this topic. I have the 14700k and z790 on the Z790 AORUS PRO X . After the update on the new microcode bios, my CPU temps are huge, also voltages are higher. I'm using an air cooler (Deepcool AK620) maybe that's also an issue. Going back to Gigabyte OPTIMIZED profile, lowers the CPU temps and vcore.

[@Xetrill]

R15 was probably schedules to only E-cores because it was deemed a long-running background task. To not reduce system responsiveness.

[@3do2]

My CP/SP Score in the Gigabyte Z790 AORUS MASTER bios with a 13900K went down with the latest bios from 88 to 80 and my CPU requests higher voltages than with the previous bios Version. Looks like a lot of people with the same issue. Does this mean the CPU is degraded ? what´s the reason to lower CP/SP Score with a new bios ?

[@gardenia1738]

I think Intel knew and knows about the high voltage, because that's what makes a lot of their crappy silicon CPUs work. It seams they have sent a LOT of bad samples to the retail market that only high voltage will make them hit their advertised specs.

[@Pacho18]

And break, over time.

[@de4ler]

CANT WAIT FOR MORE VIDEOS . TEST MORE

[@Savigo.]

What is the maxium voltage spike value compared to HWInfo readings You've ever noticed? +50mV, 100mV, 200mV? Since I don't have an osciloscope I would like to know how much voltage I should add to the vcore displayed in HWInfo to estimate the real value of the actuall highest voltage. Right now I'm sitting at stable 1,08V max underclock + undervolt 13600K - default AC LL = 0,25 mOHM DC LL = 1.1 mOHM (yeah, that's default on my MSI board and I don't want to change a value that I don't fully understand). Max VCore reading is slightly lower than VID (1,084V vs 1,099V) and It's the oposite at low loads (Vcore =0,725V VID = 0,655V).

[@md.parvezhossain1853]

microcode 0x12b update video please

[@phazeF]

I was thinking if we could use Gigabyte's optimization profile back after the microcode eliminating the VID bug requests, also note, the optimization profile sets the AC/DC LL to 0.4/0.9.

[@PAIN-ot4cj]

Should I trun of cep if im allcore overclocking on a Asus Encore z790 board with this new Microcode

[@ActuallyHardcoreOverclocking]

if you're setting the core voltage manually CEP being turned on is just gonna cause problems. So I'd turn it off.

[@PAIN-ot4cj]

@@ActuallyHardcoreOverclocking yeah I’m doing adaptive voltage, still going to wait for the final release of the bios since I’ve never had instability problems. Thank you.

[@ActuallyHardcoreOverclocking]

@@PAIN-ot4cj with adaptive CEP should work.

[@Throttler]

so basically, you do the Intel fix and then you do the BZ fix