BSOD and crashes to dekstop on many video games

[idanlevayev]

The thing is where do I find it, what is the name of the place, I don't want to make a mistake

I will try to look around the bios and see if I find it

 

[BoggledBeagle]

The thing is where do I find it, what is the name of the place, I don't want to make a mistake

I will try to look around the bios and see if I find it

I posted the page from the manual on the previous page.

Again:

 

[idanlevayev]

Hey I uploaded the photos from the bios

I am sorry I am being kind of annoying about it, its just I am not 100% sure what I am doing and if I do it correctly, do you mind telling me where and what should I change from the photos?

 

[bug]

But if it really is the problem, how come that the heaven benchmark could run on low without crashing and on higher quality it always crashes, (now it does not)
+it crashes on a specific game which is not the most demanding game I am playing.. :\

On low there probably wasn't enough stress on the GPU, so it ran fine drawing less power.

Hey just did the cinibench again after the PC rebooted,

I am pretty sure that is a problem.

Leave the voltage thing alone for the time being and start with the beginning: re-seat you CPU heatsink and reapply the thermal paste. Just like that cable, it may have become loose and not do its job properly anymore.

 

[idanlevayev]

On low there probably wasn't enough stress on the GPU, so it ran fine drawing less power.


Leave the voltage thing alone for the time being and start with the beginning: re-seat you CPU heatsink and reapply the thermal paste. Just like that cable, it may have become loose and not do its job properly anymore.

Hey, I have already re-placed the cooler and replaced thermal paste once I saw the 100C, it still did not help, what I did was setting the fan to minimum 600 RPM and changed the ASUS multicore Enhancement to "Disabled - Enforce All Limits", after doing it the temps never go higher than 70 while gaming on every game even those who usually crashes, but I still don't think it's solved, because when I tried running Cinebench it crashed immdiatly.

(cpu)the wattage while gaming averages now on 80 max was 109, it was 246 max and averages on 220, which according to BoggledBeagle who helped me a lot, is the problem, and it should max on 160 like the cooler, but I am not so sure on how to do it.

now because it barely reaches 100 (which does not crash from any game) but on R6 I get around 80 FPS, and before that I used to get around 210-240.
So I would really like to try and maybe change the ASUS multicore Enhancement from disabled to "Enabled - Remove All limits(90C)" and limit the wattage to 160 so it wont overheat and still have good performance.

 

[mechtech]

try checking event viewer, might have more info on errors etc. there

 

[bug]

Hey, I have already re-placed the cooler and replaced thermal paste once I saw the 100C, it still did not help, what I did was setting the fan to minimum 600 RPM and changed the ASUS multicore Enhancement to "Disabled - Enforce All Limits", after doing it the temps never go higher than 70 while gaming on every game even those who usually crashes, but I still don't think it's solved, because when I tried running Cinebench it crashed immdiatly.

(cpu)the wattage while gaming averages now on 80 max was 109, it was 246 max and averages on 220, which according to BoggledBeagle who helped me a lot, is the problem, and it should max on 160 like the cooler, but I am not so sure on how to do it.

now because it barely reaches 100 (which does not crash from any game) but on R6 I get around 80 FPS, and before that I used to get around 210-240.
So I would really like to try and maybe change the ASUS multicore Enhancement from disabled to "Enabled - Remove All limits(90C)" and limit the wattage to 160 so it wont overheat and still have good performance.

I would forget about multicore enhancement. Leave it off and use this: https://www.techpowerup.com/review/intel-core-i9-12900k-alder-lake-tested-at-various-power-limits/
Set the power limits yourself. Set PL1=PL2 at first, see how much is stable. And then you can lower PL1 so it doesn't run full-throttle all the time. Iirc my 12600k runs PL1=95W, PL2=125W. I'm probably not getting the peak turbo the others do, but other than that, the only difference between this and running stock is that now I can barely hear the fan spinning.

 

[idanlevayev]

Hey, should I also lower the VID voltage?
I see the VID gets everything it asks for, the vcore = vid.
and its around 1.4v -1.5v

 

[BoggledBeagle]

Hey I uploaded the photos from the bios

I am sorry I am being kind of annoying about it, its just I am not 100% sure what I am doing and if I do it correctly, do you mind telling me where and what should I change from the photos?

You need to type 160 where you see Auto, you need to see 160 instead of 4095 in both places.

Hey, should I also lower the VID voltage?

First set the power limit and fix the overheating. This is the only thing people really need to change.

Only after that if the stability problems will persist, try increasing voltage a bit (positive offset of 50mV). If it fixes the problem, your CPU is degraded and you should be able to get a replacement or refund in warranty. If you have no warranty, you will have to run the CPU with increased voltage.

 

[idanlevayev]

Hey, just did the 160 wattage limit, and disabled the multi core enhancement, ran cinebench and for the 1st time it managed to complete the 1st picture after 4 minutes.
It spiked at 94C and averages on around 86C - 88C.

But once it finished the first picture, the computer freezed, I mean it looks like it works much better now, it never ran cinebench for more then 15 seconds.

 

[BoggledBeagle]

It spiked at 94C and averages on around 86C - 88C. But once it finished the first picture, the computer freezed.

The CPU should be able to handle these temperatures, they are still high though. Try again lowering both power limits to 130W.

There is nothing especially uncommon, that a CPU may need a little bit more voltage to run properly. But when the CPU cannot run properly with stock settings it is formally defect and you should be able to get a new CPU/refund.

I asked you if you got the CPU as new or used, if used you can be never sure what the previous owned did with it and probability that it is degraded is much higher.

Running the CPU with incorrect cooler/incorrectly adjusted for the cooler at temperature limit also does not help the CPU in any way.

 

[idanlevayev]

Unfortunately, I think I will not get a refund... here where I live refund is usually not an option even tho the warranty says 36 months, I will try to ask for it, but until then should I just lower the wattage?
I don't mind lowering it, if it means it will be more stable and I can play without blue screen or turning my PC to a DIY bomb.

 

[BoggledBeagle]

You need to figure out if you can stabilise the PC with changing CPU settings, if yes and the PC will become unstable again after resetting the settings to stock values your CPU is defective.

 

[DeathtoGnomes]

IF heat is a problem try this mod...

 

[idanlevayev]

You need to figure out if you can stabilise the PC with changing CPU settings, if yes and the PC will become unstable again after resetting the settings to stock values your CPU is defective.

Alright, I will try to talk with the store I bought it from.. thanks for the help!

You need to figure out if you can stabilise the PC with changing CPU settings, if yes and the PC will become unstable again after resetting the settings to stock values your CPU is defective.

So if I don't get the refund/replaced CPU you recommend buying a new one?

 

[BoggledBeagle]

So if I don't get the refund/replaced CPU you recommend buying a new one?

You need first to be certain that the CPU is defective, and if it is, it will be easy to formulate the warranty claim. You will then receive new CPU.

I have no idea if you managed to stabilise the PC somehow, I was suggesting trying to increase the CPU voltage.

 

[efikkan]

Guys, the reported issue here is clearly a memory corruption issue, so the GPU, PSU, CPU cooler, drivers etc. isn't the issue.
The most common culprits for such problems are;
- Overclocked CPU (including running power limits above spec)
- Overclocked memory
- Defective memory (I believe you may have eliminated this one, but it's at least fairly easy with a proper MemTest86/MemTest86+)
- Degraded/defective IMC
- Corrosion on contacts (highly unlikely on a 1 year old PC)

Ram: Corsair Vengeance RGB 2x16GB DDR5 5200MHz

Can you please give me the precise SKU, as there are several variants of this product.

What do you mean by degraded? how can it happen?

All microchips degrades over time, that's why they all eventually stop working as intended. Degradation may come from productions issues, but is also accelerated by voltage, heat, and prolonged load. Degradation is why someone can achieve a "stable" OC today, but the PC crashes a few months from now.

Right... For these reasons, I always say , buy a CPU with iGPU....

Who doesn't have an old GPU lying around?

The Windows memory test you ran has two modes: quick and extended. If you already did the extended test, you're good.

The Windows memory test isn't very extensive. I'd recommend running either MemTest86 or MemTest86+ for a few hours. It's usually not required to run it for days, but any such utility will still not catch all timing issues.

For the SSD, there's SMART. Install the software from the manufacturer, it will tell you about drive health, number of errors (a small number is nothing to worry about) and general drive health.

Even though the SSD isn't causing the corruption of the memory heap (so any issue found here is a separate unrelated problem), I want to add that any SMART error should be taken with concern, as it's a sign of a pre-failing drive. SSDs are prone to becoming unstable long before their "endurance" is depleted. Running a diagnostic/SMART-test on a drive yearly is smart, probably twice a year for your OS drive.

 

[P4-630]

Who doesn't have an old GPU lying around?

I don't, many don't have that...

 

[idanlevayev]

Hello, I have a question, how can you be so sure it's a memory corruption issue? I mean the CPU is overheating on every video game and the problematic one which is R6S gets 90+ degrees.

I checked the paper with the billing info and what I bought, I will write it down exactly as it written down : Corsair Vengeance RGV 2x16GB DDR5 5600MGz CL40 CMH32GX5M2

About the memtest86, I will start running it and tell you after a few hours.

I am not sure if I have done it correctly but here is a photo of the SSD health

 

[bug]

Guys, the reported issue here is clearly a memory corruption issue, so the GPU, PSU, CPU cooler, drivers etc. isn't the issue.
The most common culprits for such problems are;
- Overclocked CPU (including running power limits above spec)
- Overclocked memory
- Defective memory (I believe you may have eliminated this one, but it's at least fairly easy with a proper MemTest86/MemTest86+)
- Degraded/defective IMC
- Corrosion on contacts (highly unlikely on a 1 year old PC)

That would be the prime suspect. But without further proof, it's not that "clear".
Tbh, memory failures are hard to pinpoint, they cause a myriad of other problems when they happen.

Running one stick at a time can help diagnosing the issue.

 

[idanlevayev]

Running one stick at a time can help diagnosing the issue

What am I supposed to look for if I do it?
Run cinebench again?
Test the games and the CPU temp?

 

[efikkan]

Hello, I have a question, how can you be so sure it's a memory corruption issue? I mean the CPU is overheating on every video game and the problematic one which is R6S gets 90+ degrees.

Your BSODs saying "PAGE_FAULT_IN_NONPAGED_AREA" and "KERNEL_MODE_HEAP_CORRUPTION" tells me something causes the entire OS memory layout to be corrupted. This doesn't happen from a bad SSD, PSU or GPU itself. Your problem has to do with CPU and/or memory.

I suspect you are running your CPU and/or memory overclocked. Start by making sure everything is stock, otherwise this will be a wild goose chase. And that includes, having the stock PL1/PL2 thresholds as specced by Intel.
Temperatures spiking to 100ºC isn't causing memory corruption (on a working CPU), it's hot, but it's not enough to cause a crash. (Now running at 100ºC sustained wouldn't be healty though, but that's a separate concern)

I checked the paper with the billing info and what I bought, I will write it down exactly as it written down : Corsair Vengeance RGV 2x16GB DDR5 5600MGz CL40 CMH32GX5M2

There are several revisions of those, but as far I can tell, all of variants I've found seem to say something like 5200 MHz at 1.25 V and 4800 MHz at 1.1 V (SPD).
If you are running yours at 5200 in your BIOS, try switching to the JEDEC/SPD profile at 4800 MHz and see if your problem goes away. This is the #1 cause of problems for PC builders today.

I am not sure if I have done it correctly but here is a photo of the SSD health

Your SSD looks perfectly fine.

Tbh, memory failures are hard to pinpoint, they cause a myriad of other problems when they happen.

Running one stick at a time can help diagnosing the issue.

Defective memory sticks are fairly easy, but other memory issues related to timings are very hard to pinpoint, and even a 48h memtest may not show an error, while another load can trigger errors after 2 min. So MemTest86/MemTest86+ are mostly to detect bad DIMMs or bad connection.
Detecting obscure timing issues due to a defective CPU or bad overclock may require a specific workload, and not always a heavy workload. I've seen OC memory pass MemTest, yet fail in compiling some core or running Prime95.

 

[idanlevayev]

I was trying to look for what you said, and saw that memory is already on 4800.

 

[bug]

What am I supposed to look for if I do it?
Run cinebench again?
Test the games and the CPU temp?

Just see if the system is stable running one stick or another. If the RAM is physically faulty, that rarely affects both sticks at once.

 

[efikkan]

I was trying to look for what you said, and saw that memory is already on 4800.

That looks good so far. Do you see the timings somewhere? I assume it should say something like CL40 - 40-40-40-77.
Then verify this against the memory profiles from a tool like CPU-Z (in the SPD tab, be aware that it will show "half" clock speeds, so 4800 MHz will be listed as 2400)