Gigabyte's Intel baseline profile runs some ABSOLUTELY INSANE core voltages.
Vložit
- čas přidán 27. 04. 2024
- My Patreon: / buildzoid
Teespring: teespring.com/stores/actually...
Bandcamp: machineforscreams.bandcamp.co...
Twitch: / buildzoid
Twitter: / buildzoid1
Facebook: / actuallyhardcoreovercl...
#intel #Z790 #14900K - Věda a technologie
Gigabyte paid for the whole voltage, so they will use the whole voltage
Gigabyte giving their customers value for their hard earned dollar.
more specifically YOU paid for the voltages, but it's good that Gigabutt is giving you the voltages YOU PAID for
I had to change the pl2 for my mobile 14900HX as my eyes nearly popped out when I saw it was allowing the processor to pull 157w and that's in a laptop!.
Gigabutt? When did Gigabyte offend you? Or do you have adorable sophmoric insults for all PC companies? 😅@imthebadguys
Electromigration speedrun
Imagine dropping the big money for a 14900k, not overclocking it at all, and having it crash on a single pass of Cinebench.
That could happen with earlier versions of enthusiast motherboards aswell. Nothing new there really. Dont buy a K part or oc mb if you dont plan to tune it. On the otherside I would be very annoyed on the motherboard manufacturer doing a piss poor job on the stock voltage loadline.
@@impuls60 Don't buy a K if you don't want to tune it? Is that seriously how CPUs should be sold?
@@impuls60you're being disingenuous. How many system integraters have shipped PCs with stock settings only for customers to have either an unstable computer or having to warranty it? This is pretty unacceptable.
Imagine spending more on a 14900KS and having to underclock it to 5.8 all core to get stable 😂 oh wait that's my reality ☹️
@@impuls60 Your comment would make sense if the non-K SKUs had the same power limits as well as base and boost frequencies as the stock K SKUs which isn't the case as the non-K SKUs are always slower than equivalent K SKU at stock settings.
I saw 1.7V and thought that motherboard vendors were high on something. Now I see that this is technically within spec so now I know it's actually Intel that are high on something
Even my Riptide Z490 with a 12600k was running 160W and 1.31V and peaks at 1.41V for single core boosts. I genuinely thought something was wrong with the mobo since that type of voltage for 4.6GHz on modern cpus is completely insane. I then did UV and it now reaches the same all core frequency and R23 score at 1.14V and 120W. And if I let it go to 160W it boosts at 5.1ghz with 1.28V.
I know that UV will always have an efficiency advantage over stock settings, but going from 1.31 to 1.14V with same clock is insane
Both? Both. Both are mad.
That would produce 528.040 Watts of heat!!! 1.72 Volts times 307 Amps = 528.040 Watts.
I guess my 360mm AIO won't suffice for that.
@@elita2cents Introducing The New 528.040mm AIO "Hells Kitchen" or "Satans Schweddy Balls Edition"™℠ (Trade Mark and Service Mark)
@@mastroitek using 13600k stock clock with 1.07v (from 1.23-1.35v); i think i5's are safe from this entire issue 😅😅
the baseline spec is fluid and open to interpretation.
"Everyone should be able to express their baseline however the floof they want" -Someone at Gigabyte, Probably
When the boss says "just set everything to maximum in this Intel document and get back to your other tasks" and the worker dgaf
Actually they are black and white. Intel has a public document stating what the profiles are. Its PL1 253w PL2 253w with voltage protections on and the correct 56sec maximum boost timer.
@@CuttingEdgeRetro You may notice that notably absent from your list are any voltages. And voltage protection should not be something you rely on day in day out.
Nein Nein Nein watts is a-ok!
When BZ said "that's not even close to the highest voltage we're gonna see today"
I went to get a brew and some snacks as I knew this was going to be good!
Intel is quite literally shooting itself in the foot without standard power & voltage limits.
AMD made this mistake with the AM5 ASUS boards, assuming manufacturers would follow guidelines. They learned fast to force limits on at least the X3D chips.
Now, it's Intel's turn lol.
they didn't force any limits with the X3D chips, they just run at lower voltages due to heat dissipation issues with the stacked cache
No, because Intel is a company that literally doesn't have feet and literally isn't an entity capable of wielding a weapon with which to shoot itself.
I believe that the AMD issue was related to XMP/EXPO. If the memory voltage was changed it impacted the SOC voltage due to a bug. Regular Zen chips could handle it, but X3D wad more sensitive.
@@cheese186 they did force limits. They have a max cap on SOC voltage now at 1.3v with new AGESA
Its a K processor. What do you expect
Intel: where moving a mouse cursor needs 30 watts
That's a dumb take. AMD idles at that
@@BGraves unless it's a threadripper or epyc (the large one), which idles at 70 watts minimum
mouse cursor power consumption should be blamed on windows imo, given that it already causes weird power spikes
@@raddysurrname7944 mouse is traveling across lots of dialog boxes causing lots of hooks in the back end to fire. you would want to run these threads at high speed on fast cores to prioritize UI speed feel. power bursts are no surprise.
@@BGraves Even in circumstances where he only moves the cursor across the cinebench scene there are power spikes. And whenever i move the mouse across my mostly empty desktop in W10, with taskmgr open, i can see usage spikes. The cursor on W95 cannot use anywhere near that amount of power (on an i486 and such) and it was responsible alright. In my opinion there's no good reason for mouse cursor to eat that much power.
Edit: also probably worth noting that the frame rate of the monitor probably limits UI responsiveness anyway
Intel wanted to break the 6 GHz barrier, no matter if all of the silicon is really up to it. I think the board manufactures has been informed by Intel, that they "should" put out "plenty" of voltage for that boost frequency.
It's cheaper than QC of all the CPUs, and set a max voltage norm.
Why not set a stock CPU boost freq a few hundred mhz lower, and have it running healthy voltages and stable?
This CPU arms race just got into mad mode.
I'm sticking to AM4. Not even AM5 has got me excited. I like what just works! R9 5950X is an efficiency powerhouse (especially if you clock/voltage it a bit down)
yeah, just -30 the 5950X and it runs pretty cool
Yeah. They pushed at the cost of Silicon longevity, voltage and stability. If you see CEP in datasheet exists proves that HW engineers knew the limits. But bean counter management team simply did not care and wanted that 6Ghz. RIP.
I'm totally agree with you, because I reduced 14900k from default 5.7ghz to 5.4 ghz using intel XTU, the temperature run at least 10°C cooler on average.
On high loads it never exceed 80°C , i prefer it these way so the hardware produce least amount of heat can last longer.
I mean they quite literally followed the specs. Intel says 1.72 max operating voltage? NICE
Jokes aside, this whole situation is getting ridiculous
Edit: another way to make it stable would be to use a fixed voltage with a sensible llc, this way the acll wouldn't even apply. That would be kind of a work to find out a stable configuration that doesn't yeet the idle voltage to 'intel baseline' values with a decent llc, but it would work in some way. The issue with that is ... people using a cpu at stock settings shouldn't have to deal with it
Edit 2: it's funny to see the gigabyte Biscuits rating changing between 9:08 (92.943) and 49:56 (92.114). Well, nobody really knows how that score gets computed and BZ clears cmos multiple times, so that's probably just variance, but heh
💯
thats for 10th and 11th gen
@@ChrisGR93_TxSwhat, the 1.72v as specs? Nah it's also for 13th / 13th refresh. Not that I would trust it, but still
@@Profetorum the shrinking of the power lines might have them starting some other max, but I am sure Intel states to OEM's that this is a max, but not anything like a max 100 of the time - Really Intel needs to just kick some teeth in. AMD learned the hard ware, and then partially forgot - they both need to learn it and never forget it.
@@xBINARYGODx yeah i mean... to me it feels like these (beta) bioses are made just as a response for the drama surrounding the intel crashing, but they are really just band-aids. A serious limitation, with some serious standards would definitely be better
12:00 While true, for cinebench purposes the 14900k isn't competing with the 7800x3d it's competing with the 7950x, which is still cheaper.
but it does perform similarly in gaming to a 7800x3d so technically you get the gaming performance of the 7800x3d with the multithreaded performance of the 7950x but at a cost of stability and power. Not worth it IMO.
Nah, its competing with the 7800x3d. I bought mine for gaming since it has the best silicon in the lineup. You get best ram and ring speed chances with the xx900 series.
Wow, dude starts off saying that he's talking about cinebench and gets two replies trying to refute it by talking solely about gaming. Good job guys.
7950x stock scores less than my shit bin 13900kf at 5.5 41k
It amazes me that intel is blaming the mobo vendors when it is the cpu's that are telling the motherboard how many volts it wants. Mobo vendors leave it open but the cpu is demanding the insane volts.
34:10 Gigabyte is undervolting on the new BIOS skin, my 12700K was 40/90 AC/DC_LL on the older BIOS skin, 13900K was 50/90, on the newer BIOS the 13900K is 40/90 AC/DC.
Good to know when it started, cheera
Around 1.5V continous is still high. I would like to see max 1.3V @continous on this size silicon.
Almost 1.7V boost is just insanity with ambient cooling.
1.35 is enough to do easily 5.5p/4.whatverE and use under 250w. And the kicker is that you still hit near 39k in C23. If you go all out on overclocking, have the big custom loops, you'll get maybe 42k, but at massive wattage increases & heat loads. For ~5% to 10% extra performance that isn't that relevant for most people.
@@bigpoppa1234 Exactly. Why it is bad that the motherboard vendors push the CPU so hard on "stock" settings. They should aim for stability/longevity on a stock settins profile. Just adding extra voltage for a tiny performance increase might lead to "stability" short term, right untill degradation.
In the old days, most of those who choose to manual overclock, knew this trade-off with the performance increase.
Most average consumers don't know this is happening, and they don't know the core temps either. High temps are not so bad if voltage is low(think laptops) But high temps are really bad if voltage is high. And the problem with electronic/semiconductor in general, is the hotter it gets, the more voltage is needed. So a vicious cycle.
The 5900X I daily drive does 1.5V SC sustained, but I suspect that's more about the boost algorithm freaking out at 20-40°c during load
@@laceflowerhw It's normal to see 1.5V during max SC boost on Zen 2 and 3. (And almost 5 GHz SC boost on 5900X) Not harmfull voltage on these shorter turbo boosts. But during sustained load, I would prefer max 1.35V on these... unless you can cool it with some extreme cooling.
Feel for you man, worst luck than me on silicon lottery lol.
Appreciate the video as always.
Gigabyte still overvolting since the days my 4790k default was 1.55v for 4.4 which i think was 4.2 allcore when I managed to get 4.4 all core at 1.15v
Luckily I was straight into the bios and noticed it before even installing an OS on the pc
The “Intel POR” profile has the values from Intel, not “Baseline”. Blame Gigabyte naming jank.
well, no. Intel POR uses AC_LL=0.4 mOhm and DC_LL=0.9 mOhm. PLs and IccMax are Intel recommended values.
@@Dakkidazeshould that be higher? Or are those values seemingly ok?
@@MonkeyMan125431 Intel **recommends** the following values for Raptor Lake-S 8P+16E models (so 13900K and 14900K):
Vmax = 1720 mV
AC_LL = DC_LL
DC_LL
Thanks for the video buildzoid! I initially got my 13900K running stable a couple of years ago on a Gigabyte Auros Master Z790 after switching default cpu loadline voltage profile from 'Low' to 'Medium'.
Your previous video mentioned limiting the tjmax to 90C instead of 100C, but you didn't mention that again in this video and focused on current limiting instead. Is the former still a good solution to prevent degradation (if that is indeed what is happening to some people) ?
So instead of the mobo figuring out a stable overclock for us, now we have to figure it out ourselves FROM SCRATCH. Thanks for making this video.
Intel Addresses Stability Issues with 13th and 14th Gen CPUs, Blames Motherboard Manufacturers😇😇😇😇
well intel certainly didn't write the bios for the Gigabyte MB i'm using that put PL1 and PL2 at 4095 watt as a default. pretty sure they specificed my chip as having a 65 and 154 watt for PL1 and PL2 respectively
I have an Aorus Master Z790 with a 13th gen core i7 on it, and noticed the same pattern as you did, even unstable at some base MB presets with a very good watercooling solution on it, it took me quite some time to figure what's wrong.
I just concluded that these intel chips are kind of "factory overclocked" since the competition is putting a lot of pressure on them, and they need that extra voltage to not crash.
And since mine is a big silicon lottery looser, overclocking is out of the question, that's the 1st time i built a system for myself and not OC the CPU.
Daily driver overclocking is kinda dead. Used to be able to overclock 20%, likely not even sustained. Best you can do on these is like 300 MHz
Hey could you test the intel baseline power feature in asus boards?, i have a 13700k with a z690-e strix and i updated the bios today to test this feature. I havent seen the vcore over 1.46v which seems fine, but it would be nice to watch a review from you on asus boards
What's the difference between the new "baseline" and the old "enforce all limits" profiles, I was under the assumption that enforcing all limits is basically the baseline, or intel advertised defaults.
Remember when people were crapping on the FX series for being space heaters? Yeah the pendulum swings both ways I suppose.
while watching this - i found an article from a couple of weeks ago that said some of the motherboard makers are distributing new BIOS with intel baseline settings
- so i went and found that there is a new BIOS for my mb ( dated 5 days ago ) - it's only new spec is the baseline setting
From my own experience motherboard companies tend to jack up vcore, my 9900k on an Asus Maximus board wants to push 1.35v+ vcore for stability on the optimised defaults setting and even higher if I do all core turbo.
With a little tweaking I got vcore down to 1.15v, which is slightly lower than Intel's default profile and dropped the temperature.
One thing i dont fully understand about the 188w power limit is that I've been getting this on my asus z690 as well and cant seem to get it to draw any more power unless doing full bios reset and allowing asus to use it's own "auto" optimized settings
Is it the same on Gigabyte z690 Gaming X mainboards? I am super spooked to change anything.
Are you using a load-line capable (line interactive) voltage regulator, like an Eaton or Tripplite UPS? I'm curious if a line interactive UPS would help with overclocking, versus a standby UPS or an always online UPS.
a UPS only gives out as the load line demands, it's not like it's blasting max amps all the time
@@erkinalp yeah but there are different types of UPS. The "online" UPS are always putting out the exact voltage, no matter what the power company is doing. The standby and line-interactive UPS have a delay
idk about the gigabyte boards but the Asus boards, you can change the cooler score and your cpu can start really pulling some voltages and running high ghz, there is so many things that will allow the cpu to just run away.
On my Aorus Elite X with my 14900K I have a similar basic BIOS setup with TVB stuff ON and ICCMax at 400 A. I also run a beta BIOS that allows to turn off the CEP feature off on my board, yet it doesn't seem to do anything in my case, so either ON or OFF seem to produce almost the same results. For the rest, I enforce PL2 at 253 W and I have way different load lines: LLC Medium, AC_LL at 12, DC_LL at 61 (which gives me VIDs = Vcore in HWinfo). This is a rather aggressive undervolt so obviously it won't work on every CPU, but IMO just raising the AC_LL a bit until stable on lesser chips would to the trick, all else being equal.
With this approach (along with XMP), my PC is rock stable (1 hour Linpack 2021) with max core temps at around 80 on a 280 mm AIO (21 C room temp), and I get around 39.6K in Cinebench R23 after one loop. My max voltage is 1,392 V at "idle" and 1,164 V during Cinebench multicore. My PL1 is set at 200 W and Tau at 56 seconds, so longer workloads (which are quite rare TBF) don't cause too much fan noise, and when I'm at 200 W my max temps are around 65 C and I still get around 37K in CB R23. No need for crazy wattages IMO, because setting more reasonable power limits from the get go allows to set better undervolts, which then result in better efficiency through more performance per watt, lower utility bills, less heat and less fan noise.
@buildzoid : What's the reasoning for the Extreme LLC setting when in another video IIRC you recommended "Low" or "Medium" for Z790 Gigabyte boards?
after watching this video... i'd RMA the CPU and Board and go back to AMD.
they probably will reject those RMAs because of them being a shitty companies
@@kingofstrike1234 they would stick four of their infamous orange arrow stickers at each of the socket's corners and call it a day. And maybe send you back your board within 28 business day...if they feel like it.
*memory trains for an hour every time you turn on the PC, blackscreens constantly because of shit drivers* - Enjoy amd bro lol.
@@bigpoppa1234 did you pray your daily papa is here already or why are you so triggered because there is a objectively better platform that is not an absolute dumpsterfire at the moment.
@@MGK195 "better" lol.
On Asus B660 its a blessing. vcore offset -0.005v and the auto voltage overblast (1.53v) set by Asus is gone. -> 34800 CB r23 @220 watts 5.2 /4.2/ 4.5 (ring) Ghz. Haven't changed anything else and for the first time this computer runs stable without fiddling around in the bios every other day. Metro Exodus Benchmark min fps went up by almost 80% and stuttering is gone in every game I tried.
Im sorry if i missed it so whats the conclusion?
I have 14900k should i just set PL1 = PL2 = 253W and 307 amps?
In a fine moment of replacing my old AIO, I bent the pins on my old motherboard with my 12900KS. The new one is a Strix Z690-E, and then a 14900K to go with it. I just got it all assembled yesterday. I think the SP on the new CPU was 93 and would finish Cinebench boosting to 6 at times, but even with a new 360 AIO, it was boiling water in no time. Then, I saw the main Vcore voltage at 1.7v. I just finished locking the voltage to 1.25, with only the P cores enabled, and at a locked 5.5. She purrs like a kitty now.
Edit: I was sustaining around 310=330w during the Cinebench run. The 3080 GPU will pull 350+ under full load. Yikes, I was closing in on the 1000w supply I have in there. I don't know what I'd be doing to fully use both GPU and CPU, but still.
I can confirm that on my new Z790 Gigabyte board with the Intel Baseline setting, the board even uses more idle power, compared to when I just put all the powerlimits to "auto", which you know, turns any limits of.
So without any power limit on my 13700K, the board uses around 21 Watts of idle power.
However with the latest beta bios, this idle power usage goes up to 31 Watts, with except for the Baseline Profile, every other setting is the same as before.
So I just down-flashed that bios back to the one before that. That baseline is no bueno.
Geez I was running these voltages 25 years ago on a 0.18 node!
I tested this on the Tachyon non-X with a potato 14900K that gets a similar Biscuits score (~93) on the F9f BIOS. R20 and R23 seem to run fine all core on the BIOS default power limit settings, but R15 not so much.
I don't see the super high volts on the Intel Baseline Profile, however.
It seems to me that Gigabyte just added a profile in the BIOS they will use as a recommendation when they get support requests for crashing CPUS. The call it "Intel baseline" but it is in fact an "overvolting baseline" so that degraded CPUs could work a few months more perhaps until they degrade even more due to the high voltage.
When it's clocked to Jupiter you need to be very cold and chunged to Saturn just to run baseline profile. Top room warmer awards goes to Intel CPU and applauded for consistency 3yrs in a row
bot?
@@xBINARYGODx no.
Intel: Gigabyte runs our processors under conditions of "sustained high voltage and frequency during periods of elevated heat"!
Gigabyte: high frequency you say? there, fixed, good thing it was just frequency
As soon as multi core enhancement was introduced by board vendors i knew i was never going to be able to trust them again, and so from that day I never run defaults on my or a client's system
High voltage I see on idle a lot. Could it be that this parameter is max voltage on the package, but a working core gets less?
I saw 1.72 vcore on a z790 strix board, not only gigabyte.
Jesus christ i thought 145v was high on mine only needs 1.32v 13900kf
What can you get out of a stable memory overclock on that x board you have?
You said that the AMD SOC 1.4v ended up getting adjusted so they couldn't set above 1.3v - doesn't that therefore mean that what they previously stated as supported (the high ram speed you mentioned) would no longer work?
The thing that bothers me the most is that Intel has been my goto CPU for people that just want a stable computer to do daily stuff.
Now, we know what's going on and we're more than happy to mess around with settings to try and fix things, in fact, we enjoy it.
But for your average Joe, they have no way (currently) to know what is happening and more than likely blame themselves for the crashes. And while we can easily change settings, Intel and board manufacturers offering a crappy baseline setting in the BIOS does absolutely nothing for a majority of consumers.
This sucks.
I have set a number of mitigating clock/power settings to stay well under 90 degrees even on small FFTs on Prime 95 on all my 13900k/14900k systems but I would certainly expect any of the CPUs to at least run Cinebench with the crazy limits. I sent 2x 14900ks chips back for that reason. If I owned the chip used for the video it would be RMA for me, even if I planned to limit everything to play nice.
Oh Gigabyte. I wonder if this is just standard Gigabyte 'oops we accidentally the whole settings', inter company politics where Gigabyte is doing something intentionally reckless to forcibly point out to intel that their spec is full of giant holes that lead to dead cpus, or if it is intel asking its board partners to bump voltage a little in the 'safe' settings to hide the degraded cpus that are out there (and gigabyte went way too far).
since intel is blaming board partners, i'd personally go very strictly with their guideline no matter how ridiciulous they would be
Shout out to madness777 👍
Crazy how a year later AM5 is actually pretty solid, I'm running DDR5-8000 with a 7950X daily and benching their new APU around DDR5-10000 while Intel can still crash in R20 out of the box and stabilizing high memory speeds feels like some kind of psychological warfare campaign.
Regardless who's fault all this is, it's absurd for a platform that has been out this long and there's no way in hell I'd run these chips at 1.60V+.
Who would have thought that stagnated silicon design and brute force performance gains through insane power consumption (permitted through a lack of real requirements) could have ever lead to this?! Oh... wait...
This is kinda unrelated, but on my 7900x + b650 aorus elite board, hwinfo regularly reports vddcr_soc voltage spikes at 1.440v when the system is sleeping despite it being set to 1.2v or lower, I've seen them go beyond 2.0v, always running latest bios at any given point, gigabyte says its nothing to worry about when I email them about it. Thanks gigabyte 👍
Sleep causes all kinds of weird sensor bugs. So there's a decent chance that those readings aren't real.
@@ActuallyHardcoreOverclocking thanks for the insight, makes a lot of sense
This is neat.
I'm having problems with my 13900K and ASUS Z690 Hero motherboard. I'm getting random blue screens. Tried reinstalling Windows, updating the BIOS and it couldn't update the ME firmware, so I had to rollback. I'm really frustrated atm! 😟
"stock bios settings" should only give the CPU the exact voltage the CPU requests, combined with intel stock PL1 and PL2.
Anything above and beyond that can be user settings or some preset 'extreme' bios setting or whatever.
Not true to be honest auto vcore will overshoot a lot usually had the exact same on my 13900kf without ai oc or mce or enhanced turbo auto vcore was over 1.45v on r23... only needs 1.32v or something to pass occt and everything... lol stock vcore will kill your shit.
@@TheBURBAN111"auto vcore" in most mobos is not "stock" per the intel cpu voltage table. it is usually much higher.
But the voltage table on the CPU is what intel has rated it to run at, so that's what it should run at *stock*.
If you wanna lower it, do it. Just don't do it out of the box. And don't raise it out of the box either.
Intel "Extreme Config" states 320W for 13th and 14th gen 8P + 16E, so what's the problem with 150W for 4 core Cinebench? You need a VERY GOOD AIO for 14900Ks, preferably custom loop or MoRa 420 (I use the latter) to just keep the temps under 100C.
Quite absurd when stock settings aren't stable. It's not the job of the buyer to tinker around just to get something to work properly out of the box.. System building has never been so unbalanced. If a system builder is piecing together a PC for someone who needs to do intensive tasks like video editing, they now have to worry about the customer's computer not functioning properly if something like a power outage resets their bios? Regardless of who is at fault, this is certainly not cool.
i got a cyber power pc with 14900k and had to underclock as u said unreal engine and all should a ask them to take cpu back and if so what should it be replaced with
I'm reminded of the 1GHz Pentium 3 that Intel ended up having to recall. This silicon is not (reliably) capable of running at the specified power and voltages.
Gigabyte also has some real shenanigans on AM5 platform, like b650 boards that have overclocking and PBO in BIOS do nothing.
Here's the list:
Gigabyte B650 UD AC
Gigabyte B650M D3HP
Gigabyte B650M D3HP AX
Gigabyte B650M C V2
Gigabyte B650M Gaming Wifi
Gigabyte B650M D2H
Gigabyte B650M S2H
Gigabyte B650M H
If you have 7000-series AMD like I do, set voltage curve to negative 20 to negative 24 (I had to disable integrated graphics to get it stable), then use a larger cooler you ever believe you need (I have 480mm water for 7950X) and tune memory. I do not use PBO, never have. I did try Asus AI overclock, but still get better overall performance by leaving other stuff to Auto, negative voltage curve, plenty of cooling and memory tuning.
AM4 too. My 5800x crashed with gigabyte x570 on default auto settings constantly until i changed everything to standard and turned off some settings. Took me more than a year to make it stable.
@@Toralian89 Are you sure your memory simply is not compatible and you got it working by setting it to non-XMP/non-EXPO, aka Jedec-setting.
Also did you upgrade BIOS?
Could also be that either your motherboard or CPU is simply defective or power supply is of low quality (wattage or being "Gold" goes not mean quality)
@@TechTusiast I'm not going over standard tech support 1.0 questionaire. It works without crashes now - it doesn't work on gigabyte's motherboard default settings. That's all there is to it.
@@Toralian89 Ok and you surely know motherboard does not dictate memory settings, so if memory is not compatible, motherboard can not help either way.
my bclk is not 100.00 no matter what.. on windows it says its 100.00... but in the bios menu it stays really close to 100 (99,80-100.20).. full amd build..
can it be the bios update, cuz it doesnt show my ram voltage aswell?
Asus maximus boards also have got vr out sensors btw
You even had XMP disabled on first boot. Though I didn't see if the Intel Baseline enabled XMP
Is it possible that Gigabyte's "Intel Baseline Profile" is also reducing the ICCMax amps to a lower setting which could explain the higher volts? Just speculating here. @ActuallyHardcoreOverclocking
Does this same behavior show up on a different motherboard from a diff vendor
My original X570 Gigabyte aorus ultra board came out the box with a max voltage of 1.5 v core.... on what was at the time a 3700x a couple weeks after launch. The board also would never run near top end memory regardless of the cpu i threw in it.
I noticed that with my son board an X470 and with the 2600x. Now with the 5080x3d in there, everything is ok, using PBO -30. If i´m not mistaken, now the max voltage is about 1.1 something volts, playing RF2/AMS2 etc, in VR, using an 4080S with HP G2 at 100% resolution using open composite.
Isn't Gigabyte the one loading the Intel VID failsafe with that setting, ignoring the chip VID and basically loading what is considered the most safe (highest) voltage for the frequency?
In the future. Intel still uses a lot of power.
I solved this "problem" by setting the vcore voltage to adaptive and the AC LLC to 10 and DC LLC to 90 or even leave it at stock and put static 5.6GHz core clock and be golden.
Once you reboot the first time it will retrain llc and over time voltage will v latch down... But clear cmos will revert that, so it's irrelevant after 1st reboot
Wat
@@vyor8837 on first boot, mobo will load llc 7-9 to make sure it's stable.. after a few reboots or some use and reboots the llc reverts to best setting, and vlatch is the minimum stable v for any frequency.. aka trained svid
Why is your clock set to the year 2044?
I managed to get my 14900K running at 1.15v 5.4GHz during a full core load with CB R23. -.100v offset.
Funniest thing is that if you just disabled hyper threading it becomes stable, less power and thermals. The only caveat is that the performance is similar to the intel baseline profile in the multi threaded workloads, but it's better in games than stock
I agree, and if you reduce the number of E-Cores to 8, it's even better for gaming.
Could you pls elaborate how limiting E-cores makes it stable? In that case would it make sense to choose 14700 instead? Aksing as I will eventually upgrade my 13600KF (runs on a regulab B760 Asrock board)
You can also disable 16 E-cores, Hyper-threading and 4 P-cores and get an overpowered version of 2500K. But no one will do this because it's stupid
@@lexkoal8657 I'd do it, mostly because I'm an idiot, but I'd do it.
@@lexkoal8657 you don't need more than 8 threads for gaming atm, so disabling hyper threading and e cores will avoid any thread allocation errors and give you far better thermals and stability. so for strictly gaming/ single threaded workloads that is the optimal configuration
It will be a cold day in hell before gigabyte gets their mobo bioses in order, theres always some setting fucked up.
I'm using a Asus Z790 Hero with a i9 13900k and I am scared updating my bios. I have no issues with my video-editing & gaming PC and will leave it at that. Could you do a video for the Asus Motherboard?
You should never update bios unless you have a good reason to.
@@Akkbar21 Great advice. So I think I stay on bios 1904 for my ROG MAXIMUS Z790 HERO.
i want to see 15900k!
It was reported elsewhere that Gigabyte's "Intel Baseline Profile" set current limit to 249A. Is this what you observed?
The fact that Asus and Gigabyte "Intel Baseline Profiles" are different, and both violate specs in different ways would seem to me to be evidence that these settings are not coming from Intel.
gigglebite just doing gigglebite things
I dont have the Intel baseline setting but something that's called Intel pom
and this gives me VR VOUT of like 1.45 in idle???
wow. I had no idea voltage had gotten so high again. I thought we'd be down in the 1.1 to 1.2 range now since it seems to drop with the node shrinks. I mean, heck, my Xeon E5-1660 V3 (Haswell-E) would not likely last long at those voltages (I already burned thru one - hehe). This current one is running just 1.275V (core) and 1.8V (VCCIN).
Interesting. What voltage killed your first cpu? I am still using a 4790K in my second computer. Currently at 4.4GHz 1.12v and 1.65v vccin. 4.7GHz 1.27v 1.7v vccin is possible but hard to cool and not much performance uplift, both settings should be fine in terms of vcore.
@@ronnie3626 I think I was at 1.875 vcc and 1.375 vcore. It was actually the uncore that went out on my 5960x tho. It was presenting like memory errors but when I would significantly declock the uncore, the stability would go up. I think it was running 44x and 32x uncore/cache, but it's been a while so I don't recall for sure.
My 12th Gen i7 on an Asus prime z690-a never has any problems. Never had a single issue with all PL
Maxxed.
7800X3D is ~ 7000pts on R20 stock, 7500+ is a very good score.
I really think the best long-term solution is for Intel to add a second LLC that has temperature as an input.
I know this is Intel and all that, and AMD made the mistake with AM5 and ASUS boards ... but pls tell me that Gigabyte AM4 B550 boards are fine (planning to get one in a few weeks).
I've never heard of AM4 having this issue at all.
Am4 is a rock these days. Go hard
i find it funny that we have to go back to 130nm P4s to get a similar max vcc(1.75v)
I dont understand why Motherboard OEM's dont include specific CPU profiles which the user can select, i.e. if the user has a 159000K cpu then the board has a profile for that specific CPU.
i have asus z690-e - and intel i9-14900k
setting for ai to oc everything - it goes to 6 ghz and vcore is 1.4 + - but it crashes BSOD frequently
tried several ways to get best speed and performance with less frequent BSOD ' s
finally - set all p cores to 57 multiplier and e cores to 42
set vcore voltage cap to 1.3 v
runs without BSOD - i have no problem running call of duty
aida64 stress test - vcore runs 1.279 and lower - cpu package power 130 - 170 watts - package temp 66 c
however
when i run cinebench 23 - at the time of selecting multi-core test -> it either crashes with that error that you got or it just crashes to desktop with no message at all
-
since i can play cod warzone with no problem and plenty of cpu power - i don't care about cinebench running
Thanks... my binning seems to be as bad as yours. But limiting current to 350A and load line to 75 seems to have fixed it. And on this setting I can retain 300w power limit and allow all cores to boost to 6ghz without this being restricted to 2 cores, so no performance losses unlike this awful intel baseline setting.
12:53 wtf? 1.72V is ridiculous, how can that be "in spec" lol.
I find it odd that none of the initial reviews caught this problem. All of a sudden it's problem.
That is the odd thing.
Most of the reviews I've seen don't have CPUs as bad as mine.
tbf thats 307A EDC ... so its pretty resonable that you will never actually sustain that even under a "heavy" load. if you could scope it the current draw would be all over the place but not exceeding 307.
if it were 307A TDC then power should be roughly I*V.
does make me wonder if the 1.72V is only for Vout at the regulator to ensure that voltage inside the core is what the vid table specifies after droopage from really bad vrm... cause man its alot of current and 200mV of droop isnt that unreasonable right
The same issue occurred with the AMD Gigabyte Master bord, after a while nothing would operate at any speed for the money you paid.
According to Steve on an H100i, pump speed didn't matter. Like no difference in temps from quiet to performance pump speed.
18:50 prediction: keep calm and raise vcore
i don't know how exactly it went with AM5, but limit was ~1.4? maybe 1.45V
problem with burning sockets was mostly Asus, due to using that 1.4 limit, while others went for 1.2-1.3V, also Asus board wasn't suppling only 1.4V, it was more like 1.5-1.6V, so power limiting on Asus side went bananas there, other than applying max recommended voltage. I think even them reducing it to more like 1.3V ended with 1.42V on probe
problem with gigabyte was that OC was saved as default, so reseting to default made it keep OC and CPU was burning, so it was way more uncommon for casual user to trigger
it might have been AMD fault, but really, since derBauer found out that 3d cache was burning quite easily, then 1.4V was recommended limit for any 3d cache part, i think even before AM5 x3d parts came out
for am5, start with all core -25pbo and increase it until no longer stable, most packages should do -30 just fine, good ones all core -50 even
As ASUS was running the CPU out of spec, it was not AMDs fault. They just learned they cannot trust the MB manufactures and forced the 1.3V limit in the AGESA.
why the mouse spasticity?
15:00 That reminds me of Windows 98, when you move the mouse the computer ran faster.
that was an interrupt starvation problem
I don't know what Gigabyte is doing recently...My 13900KS runs 1.409V VR Out max. with single core load 6GHz on Apex Encore and never had any issue with the CPU or instability and never had to use any baseline profiles etc.
If you run any kind of Air cooling try this if you are having issues "Gigabyte Elite AX-W Z790 : Set PL1 to 253 PL2 to 125 , amp 307 "ICCmax" , turn off all Enchaned multicore and Enchaned TVB , and choose load line "Normal" you lose like 8-10% performance but in my case the most of the time I don`t see over 85c anymore might go to 90c in very high load benchmark. Already lost one CPU I9 14th K due to the bad settings on the motherboards.
TIL what a CPU Biscuit is. Yum!
Makes me wonder how these cpus even made it through validation