News New firmware addresses Intel CPU gaming instability — Asus Intel Baseline Profile fixes crashing on Raptor Lake and Raptor Lake Refresh chips

[CmdrShepard]

Some people have faith...

It would be far better for them if they stopped relying on faith and luck and started learning how to actually do things.

Will actually configure and design things to work safely out of the box and let us configure if from there.

That's exactly how all the mainboards are designed to work. They boot using safe-defaults and ask you to enter setup to configure them. Did you ever build a PC yourself when you don't know this is how it works, or are you just spewing unfounded anti-corporate rage and BS out of your rear end?

You have some skewed view of what unlocked means.

Not really, I am using a dictionary and unlocked there still means the same as it did for the last couple of centuries.

Unlocked means safe factory settings Aka a baseline, to tweak off of.

No it doesn't -- unlocked means there are no limits enforced by default on settings such as voltage, multiplier, package power limits, and current. So when the CPU boots, the default state of the MSR registers controlling those settings is no limits. The only difference with voltage is that it's controlled by VID and hardware sensing but that can be overriden too.

Hitting accept to modifying non-typical BIOS settings once isn't going to hurt you.

Until those ignorant people whose freedom to be stupid you are defending so valiantly say that they have accidentally clicked it without reading and then vendors will change it so you have to click it every time you enter to change anything. Some BIOS vendors are already doing this and it pisses me off to no end, that's why I am so mad about it.

 

[thestryker]

Yes, because everyone else is ignoring it and it should be addressed.

Yes, those mainboards and K CPUs are marketed towards people who know how to manually change settings.

But please, do enlighten us, what would the mobo makers have to do for you to consider yourself informed?

Because, as far as I know (and I know a lot on the subject), the first time you insert a new CPU in the socket and boot the system the first message you see is along the lines of "It looks like you have just installed a new CPU, press F1 to enter Setup to configure it".

And when you do it's all there black on white -- it just needs someone who knows how to read.

Also, for at least a decade or so, the BIOS-es have safe defaults they use to boot your CPU and run BIOS itself, and your O/C settings kick in only when the OS starts booting.

So, until you enter setup and either confirm or change default settings mobo makers don't use "whatever settings they want without informing you" so please stop spreading FUD.

I'm curious do you think it's reasonable for motherboard manufacturer's default settings to be anything other than Intel stock?

I may be wrong, but I'm pretty sure sticking to stock for default is what most folks around here are advocating for. Here's an example if somehow you're not aware of the state of things:

Most of these boards have no limit for what Intel calls PL1, a temporary power state. After a certain period, around a minute typically, PL2 kicks in, which is a reduced power state. The Asus and Asrock boards revert to a 253-watt power limit, while the Gigabyte boards limit to 280 watts. MSI, on the other hand, has no power limit for PL1 or PL2, and Asrock has decided to do this with their Nova board, but not the Riptide

taken from the written version of the HUB roundup: https://www.techspot.com/review/2776-intel-z790-motherboards/

Also to nitpick MSI they really need to ditch tying power profiles to cooler types. Though none of those profiles are Intel stock anyways.

 

[SyCoREAPER]

It would be far better for them if they stopped relying on faith and luck and started learning how to actually do things.

That's exactly how all the mainboards are designed to work. They boot using safe-defaults and ask you to enter setup to configure them. Did you ever build a PC yourself when you don't know this is how it works, or are you just spewing unfounded anti-corporate rage and BS out of your rear end?

Not really, I am using a dictionary and unlocked there still means the same as it did for the last couple of centuries.

No it doesn't -- unlocked means there are no limits enforced by default on settings such as voltage, multiplier, package power limits, and current. So when the CPU boots, the default state of the MSR registers controlling those settings is no limits. The only difference with voltage is that it's controlled by VID and hardware sensing but that can be overriden too.

Until those ignorant people whose freedom to be stupid you are defending so valiantly say that they have accidentally clicked it without reading and then vendors will change it so you have to click it every time you enter to change anything. Some BIOS vendors are already doing this and it pisses me off to no end, that's why I am so mad about it.

You live I'm some alternate reality so I'm just ignoring you moving forward. No need to waste my time reasoning with the unreasonable.

 

[SyCoREAPER]

I'm curious do you think it's reasonable for motherboard manufacturer's default settings to be anything other than Intel stock?

I may be wrong, but I'm pretty sure sticking to stock for default is what most folks around here are advocating for. Here's an example if somehow you're not aware of the state of things:

taken from the written version of the HUB roundup: https://www.techspot.com/review/2776-intel-z790-motherboards/

Also to nitpick MSI they really need to ditch tying power profiles to cooler types. Though none of those profiles are Intel stock anyways.

Absolutely.

Also I can attest to MSI going nuts with stock power limits or lack there off. Part of why my 13700 started to bake very quickly.

 

[Pierce2623]

I knew /posted this was a BIOS issue, sure enough here we are. Also surprise, ASUS is the main culprit once again. So glad I didn't go ASUS, they've been screwing motherboards up for the past 2 or 3 years.

Asus isn’t the main culprit. Every z790 board sold defaults to unlimited power pl2.

 

[ohio_buckeye]

I was watching a video by Jayz2cents earlier and he was showing how that boards tend to apply settings that are higher than intel’s limits.

One point he did make though, it could be Intel was afraid of being behind amd that perhaps they allowed CPUs to be pushed to the bleeding edge of stability.

Another point I agree with, is that the Intel safe limits should be the default out of the box settings on the motherboards. Then they should give a manual with documentation to say if you want more performance here’s how to accomplish that but here are the risks and that going to this extra performance profile etc is at your own risk. I think he was using a 13900ks and he showed how he changed settings back from the board default settings to Intel defaults. The voltage dropped etc and the temps with the aio went to the 70s in his cinebench run vs the 89 or 90 it was pushing.

 

[SyCoREAPER]

Asus isn’t the main culprit. Every z790 board sold defaults to unlimited power pl2.

That has been established had you read further.

 

[thestryker]

One point he did make though, it could be Intel was afraid of being behind amd that perhaps they allowed CPUs to be pushed to the bleeding edge of stability.

I've seen this commentary before, but it is rather unlikely because MCE has been around in various fashion since Ivy Bridge and Intel has never told them to stop.

 

[TerryLaze]

One point he did make though, it could be Intel was afraid of being behind amd that perhaps they allowed CPUs to be pushed to the bleeding edge of stability.

I've seen this commentary before, but it is rather unlikely because MCE has been around in various fashion since Ivy Bridge and Intel has never told them to stop.

Also this point falls apart if you look at actual real power draw and not the stress test numbers. At intel stock turbo settings (253W ) the 14900k is 30W above the 7950x in power draw at the same performance, in this example in blender which is pretty high in power draw, while that is lower efficiency it's not the huge difference everybody is trying to convince people with by only showing numbers from prime95 with everything unlimited....
Techpowerup

 

[CmdrShepard]

I'm curious do you think it's reasonable for motherboard manufacturer's default settings to be anything other than Intel stock?

I think that default settings can be (and apparently are) different depending on which type of mainboard you buy (basic models with non-Z chipset .vs. high-end Z chipset mainboards for O/C), and which type of CPU you plug into it (non-K SKU .vs. K SKU).

As an example, non-K CPU and K CPU can have the same default value of 0 in some MSR (model-specific register).

If the BIOS doesn't write said register at all to set a different value, then the CPU will use Intel defaults which may not be the same between K and non-K models.

On a non-K SKU the value of zero might mean "use default fused power limits", while on a K SKU it could mean "use no limits". That's why it's called Model Specific Register after all.

So what I am saying is that by buying an unlocked SKU, you are the one who has chosen the defaults to have no limits. That of course doesn't mean the CPU will burn out because it has safeguards -- it just means that unless you also have a proper power supply, beefy cooling, and high-end motherboard with good power delivery your K SKU is going to be unstable.

Why is this so hard to grasp?

It's the same thing as buying a Porsche which can accelerate from 0-100 km/h in 3 seconds and reach 300 km/h speed .vs. buying a regular car which has 150 km/h top speed. When you choose a Porsche it is assumed you know what you are buying. Moreover, it's your own responsibility not to floor the pedal and to ensure you have adequate length of empty road ahead of you if you do.

Go ahead and tell me why do you think buying high-end CPUs and mainboards marketed to enthusiasts shoud be any different than buying a freaking sports car?

I may be wrong, but I'm pretty sure sticking to stock for default is what most folks around here are advocating for.

All I am saying is that "no limits" IS THE STOCK / DEFAULT FOR K SKU.

It's by design (after all it's sold as unlocked), and this limiting that has been added recently in BIOS is a knee-jerk reaction from vendors because as usual the press has blown the problem out of proportion and blamed vendors for not using Intel defaults (when Intel defaults are "unlocked") instead of a handful of people who bought stuff they should have had no business handling without supervision being so computer-illiterate that they don't even know what "unlocked" means let alone how to properly configure it themselves.

 

[TerryLaze]

So what I am saying is that by buying an unlocked SKU, you are the one who has chosen the defaults to have no limits. That of course doesn't mean the CPU will burn out because it has safeguards -- it just means that unless you also have a proper power supply, beefy cooling, and high-end motherboard with good power delivery your K SKU is going to be unstable.

Why is this so hard to grasp?

What you are saying here has nothing to do with what's happening!

Here look at that, 11 mobos = 7 different default settings for the exact same CPU, and only 4 mobos that are close enough to the default 253W to call it default.
Mobo makers just use whatever settings they want to use, it has nothing to do with what the CPU tells them.

View: https://youtu.be/XL08mmE0RB4?t=916

 

[thestryker]

All I am saying is that "no limits" IS THE STOCK / DEFAULT FOR K SKU.

So in other words what you're saying is that you don't know what you're talking about at all, but are sure that you're right when you're not. Got it no reason to waste more of my time then.

 

[CmdrShepard]

Here look at that, 11 mobos = 7 different default settings for the exact same CPU, and only 4 mobos that are close enough to the default 253W to call it default.

So what you are saying is that the user isn't and shouldn't be responsible for the choice of motherboard and CPU they have bought?

And you are saying that they aren't and shouldn't be responsible for entering setup and changing / confirming at the very least those same CPU settings as they are asked to do by the BIOS prompt?

Because if that's what you are saying I call BS.

I have been using unlocked Intel CPUs and high-end mainboards since Z170 chipset and I am using an unlocked SPR Xeon right now. I have never, ever had any stability issues with my CPUs past or present.

Perhaps I am doing something different then? Something right, while those whiners are doing something wrong? If they can't figure it out they should pay someone to do it for them or just give it up and choose a different hobby.

Mobo makers just use whatever settings they want to use, it has nothing to do with what the CPU tells them.

The video you linked is showing just the thermals within the preset power limits and it does so without going into any details about CPU power management settings of each mainboard or even CPU voltage used.

It also doesn't show the actual CPU performance, and it doesn't show what happens with non-K SKU which makes the whole comparison utterly pointless.

It can only answer a single question "on which mainboard by default our CPU sample runs the coolest" and that's it -- it doesn't seem to explain why is that happening, nor what other settings are different nor it goes into details about power delivery stage of each mainboard.

Gigabyte is consistently coolest even without power limit, what does that tell us? Is it the CPU settings they used? Or is it superior mainboard trace routing and component layout around the CPU socket and the socket / backplate material which diffuses heat better than the competitors?

Unfortunately, we will never know because stirring rage for clicks is far more important.

In my current mainboard BIOS it says next to RAPL settings:

Use this feature to configure PL1 power limit. If the value is 0, the fused value will be programmed. The default is 0.

What that means is that my BIOS reads the fused CPU power limits from one CPU MSR and writes them into the other MSR that controls the actual power limits.

For CPUs that aren't unlocked entering higher power limits than the fused ones won't work because CPU will just ingore the values written into the control MSR -- writing higher power limits than the fused ones is only possible on unlocked CPUs which brings us back to the user choice.

If they have chosen non-K SKU the CPU would have ignored higher power limits set by the mainboard BIOS (or the BIOS wouldn't even try to change them in the first place).

But I guess it's easier to blame manufacturers for setting a 10% higher limit (or in some cases no limit) on an unlocked CPU than taking responsibility for your own ignorance and the consequences of your actions resulting from it.

 

[35below0]

Perhaps you are doing something different. I have a K intel and a non-high end z790 motherboard and it's not cooking. Or unstable.
Granted, it's an i5 not 7 or 9.

As for user choice, that argument is completely flawed. Unless a customer is sold a CPU that will be overclocked beyond manufacturer defaults, unstable, and uncontrolable. Who would buy such a CPU? And if you are going to demand that everyone wanted to be prevented from having their CPU pushed beyond what is stable, go ahead. If motherboard manufacturers set limits higher than intel's default, it cannot be user choice. It also cannot be user choice when the user is not told how the motherboard is set when it comes to CPUs.

Really, it also cannot be user choice when non-K versions of some intels sell for a higher price than K models.
Checking PPP US right now (lowest available price):

12600K - $177
12600 - $273
13600K - $273
13600 - n/a :/
13700K - $344
13700 - $371

So tell me. Which one would you pick? Only the 13700 is reasonable of the ones i listed, because the K would need a $30-40 cooler. Even then they're the same price.

It doesn't have to be installed onto a z motherboard though. But choice of motherboard never involves likelyhood the CPU will be overpushed. Usually it's the opposite, and buyers worry whether the VRM will be good enough. On budget boards, it's sometimes feeble for i7s or i9s. Budget boards also don't have all the features a flagship chipset can offer.
So how can you argue that buying the best, the flagship, the most expensive and most capable is somehow also a known and accepted risk? Customers look at features, ports, durability, and above all price. And reviews. No review i've ever seen said don't buy this expensive motherboard, unless it was because a budget version offers similar features for a lower price.

There are extreme overclocking motherboards. And there are i9 KS models. Had you argued about people who buy and overclock those it would be user choice to push limits, but it's again the manufacturers who let the customers down. Or made an honest mistake and pushed too far.


The core of the issue is that motherboard makers know they can safely exceed intel's recommendations, and so they do. This would probably continue to fly under the radar were it not for recent instability issues. Issues that would not happed had they not gone too far.
They will probably back off with the next crop of motherboards.

Nothing here is the fault of the user.

But I guess it's easier to blame manufacturers for setting a 10% higher limit (or in some cases no limit) on an unlocked CPU than taking responsibility for your own ignorance and the consequences of your actions resulting from it.

Nonsense.

 

[SyCoREAPER]

@35below0

Some of us drink coolaid while others prefer bleach. The bleachboys went out of style decades ago, wait that's the beachboys. Drinking bleach is always bad.

To the point, some just can't be convinced to exit the box.

 

[TerryLaze]

So what you are saying is that the user isn't and shouldn't be responsible for the choice of motherboard and CPU they have bought?

And you are saying that they aren't and shouldn't be responsible for entering setup and changing / confirming at the very least those same CPU settings as they are asked to do by the BIOS prompt?

Because if that's what you are saying I call BS.

Huh?! That's what I'm pointing out!
You have to know what you buy or you have to know how to change your bios, otherwise you might get stuck with settings that you don't like.

You said that settings come from the CPU they use and that video clearly shows that to not be true.

The video you linked is showing just the thermals within the preset power limits and it does so without going into any details about CPU power management settings of each mainboard or even CPU voltage used.

It also doesn't show the actual CPU performance, and it doesn't show what happens with non-K SKU which makes the whole comparison utterly pointless.

It can only answer a single question "on which mainboard by default our CPU sample runs the coolest" and that's it -- it doesn't seem to explain why is that happening, nor what other settings are different nor it goes into details about power delivery stage of each mainboard.

People with eyes have a clear advantage here..
Within the bar is the power draw that each mobo gets out of the cpu as well as the clocks they get.
Default mobo settings are all over the place.

For CPUs that aren't unlocked entering higher power limits than the fused ones won't work because CPU will just ingore the values written into the control MSR -- writing higher power limits than the fused ones is only possible on unlocked CPUs which brings us back to the user choice.

If they have chosen non-K SKU the CPU would have ignored higher power limits set by the mainboard BIOS (or the BIOS wouldn't even try to change them in the first place).

Power limits are not part of any fused limits, all the cpus can have more power, that changed years and years ago.
Locked cpus will still use a lot less power since they can't actually increase clocks above the limit, but basically their turbo limits are so high that it wouldn't make any difference anyway.

 

[InsGadget]

How would shader compiling influence any windows files?

This issue affects a lot more than just shader compiling for Unreal 5 games, despite the article's focus. I am running into crashes multiple times a day while content creating in various Adobe and other apps, on top of some occasional games crashes when CPU-bound. i7-13700KF on Asus Prime Z790M-Plus.

Installing this new BIOS now and hoping to put this problem to bed.

 

[bit_user]

This issue affects a lot more than just shader compiling for Unreal 5 games, despite the article's focus. I am running into crashes multiple times a day while content creating in various Adobe and other apps, on top of some occasional games crashes when CPU-bound.

Someone else commented on one of these articles that they experienced crashes compiling code with GCC. It does seem to affect a diversity of workloads (I think all of them multi-threaded ones).

 

[vMax]

This issue affects a lot more than just shader compiling for Unreal 5 games, despite the article's focus. I am running into crashes multiple times a day while content creating in various Adobe and other apps, on top of some occasional games crashes when CPU-bound. i7-13700KF on Asus Prime Z790M-Plus.

Installing this new BIOS now and hoping to put this problem to bed.

A couple of points. How has the new bios worked? Jay's Two cents reported that even with the new bios Asus were still a bit fast and loose with terminology on the settings.

Also I would run HardwareInfo, the latest version and start your content creation workloads and monitor through HWinfo the cpu speed, temps, mos temps and vcore to see how it is boosting, what max temps you are hitting and the max voltage under the workload. If you have not overclocked, this could be some other issue related to drivers or a software conflict through Windows.

I would also manually set vcore to no more than 1.240v as the 13700K needs njo more than that to run at max boost and speeds. This will make a big difference to temps!

Also good to know what cooler you are usiing. FYI I am running a 13700K with a mix of prod rendering workloads including Adobe, DaVinci and gaming with a fixed vcore at 1.245v with no stability issues at all though my motherboard is of a much higher spec and I am using a 360mm AIO, though I do overclock and undervolt.

 

[ohio_buckeye]

I think this is the video you are referring to. I think a little way in he gives a fix.

View: https://youtu.be/D3StcUhVRWQ?feature=shared

 

[InsGadget]

How has the new bios worked?

Also good to know what cooler you are using.

So far, so good. Currently rendering a UHD h.265 video in Premiere Pro with software encoding for the past 15 minutes. All CPU cores are basically maxed out, but temps are staying around 90 C. It would hit 99 before this new BIOS, and it would definitely have crashed by now.

CPU cooler is Cooler Master Hyper 212 Halo, which just barely fits inside my micro ATX case. Yes I should probably get a bigger mobo, case, and cooling for this CPU, but this seems to be working. I am happy again.

I'll report back if that changes.

 

[35below0]

CPU cooler is Cooler Master Hyper 212 Halo, which just barely fits inside my micro ATX case. Yes I should probably get a bigger mobo, case, and cooling for this CPU, but this seems to be working. I am happy again.

The cooler is not the greatest, but you know this. It doesn't matter because space is limited and it can still do it's job.

Your cheapest and bestest option at that size is the Thermalright Assassin X 120 Refined SE, if it can fit.
It's 148mm tall so it should. *should 😉