Is my CPU slowly dying?

[no_1_dave]

I bought my 13900k close to their release and managed to tame it by simply putting a -0.05v offset on the vcore
This allowed me to get over 40,000 in Cinebench R23 with a max of roughly 250-260w holding a temp of around 81c

Fast forward all the bios nonsense regarding stability, mine was always stable so I ignored that and didn't want to install a bios which limited my clocks etc
Approx 6 months ago I started getting the odd BSOD when hammering the CPU so I reduced the vcore offset to -0.03v and all was well again

A few months back I upgraded the bios because I got the new 50 series GPU and decided it would be best I update everything, and wow. The new bios limited performance in a huge way, sinking my R23 score down to 34000.
Hopped into the bios, removed all the new limits Intel put in place and set the offset back and I'm back at 40,000 R23

Now I've started to get the odd crashing and have had to reduce the offset again, this time down to -0.02v. So it seems I am slowly having to allow the CPU more power to maintain performance
Windows logs show the crash to be Event 19 - WHEA-Logger which points to CPU. It doesn't BSOD, but the system will just completely freeze so I need to hit the reset button. But allowing more voltage has fixed it.

So, is the clock ticking on my CPU?
Is there anything I can do about it without nerfing my performance?

Are there any cheap upgrade/side-grade options? Bare in mind I don't only game, I encode x265 as well which is why I've never considered swapping to an X3D chip

 

[ED209]

If you think the CPU is dying then RMA it to Intel. No point doing any side upgrades and spending money you most likely don't need too.

It does sound like its degrading more and more, What do you use the PC for?

I assume the PC in question is the 1 in your sig? (Gigabyte Z790, 13900k, 64GB 6000 CL30, 5090 FE)

 

[Qitarah]

As the above says, you should have a few months warranty on the CPU left? They came out in Q4 22.

Then again if it runs fine at stock settings and you cant induce any crashes with pushing it, I don't know if an RMA would be accepted?

 

[ED209]

Intel are extending the warranty for another 2 years on certain 13th and 14th gen processors so should still be in warranty due to the issues with these processors.

I would stick the processor back to stock and test, if it still has issues then RMA it.

Intel® Boxed Processors Three-Year Limited Warranty Terms and...

Three-year limited warranty information for Intel® Boxed Processors in English and other languages.

www.intel.com

 

[no_1_dave]

If you think the CPU is dying then RMA it to Intel. No point doing any side upgrades and spending money you most likely don't need too.

It does sound like its degrading more and more, What do you use the PC for?

I assume the PC in question is the 1 in your sig? (Gigabyte Z790, 13900k, 64GB 6000 CL30, 5090 FE)

Yes, it's the one in my sig. It's used for gaming, but I also regularly set it to do encoding overnight and while I'm at work

Intel are extending the warranty for another 2 years on certain 13th and 14th gen processors so should still be in warranty due to the issues with these processors.

I would stick the processor back to stock and test, if it still has issues then RMA it.

Intel® Boxed Processors Three-Year Limited Warranty Terms and...

Three-year limited warranty information for Intel® Boxed Processors in English and other languages.

www.intel.com

At stock it is absolutely fine, even with the bios set back to how it was on release where it will allow 280w. The problem is that 280w causes temps to hit 90+ while I'm encoding (using a 360 AIO), so having the -0.05v offset brought the temps down and actually boosted performance as a result. But now the best offset I can set is -0.02v meaning temps are getting to 85c. -0.05v is no longer stable at all and I'm slowly having to remove the offset back towards stock volts to hold it steady. I get no crashing or BSODs at -0.02v, but I used to get no crashing at -0.05v, then -0.03v, onto -0.025v and now I need -0.02v. It won't be long before I will need to remove the offset altogether it seems

 

[ED209]

Yes, it's the one in my sig. It's used for gaming, but I also regularly set it to do encoding overnight and while I'm at work



At stock it is absolutely fine, even with the bios set back to how it was on release where it will allow 280w. The problem is that 280w causes temps to hit 90+ while I'm encoding (using a 360 AIO), so having the -0.05v offset brought the temps down and actually boosted performance as a result. But now the best offset I can set is -0.02v meaning temps are getting to 85c. -0.05v is no longer stable at all and I'm slowly having to remove the offset back towards stock volts to hold it steady. I get no crashing or BSODs at -0.02v, but I used to get no crashing at -0.05v, then -0.03v, onto -0.025v and now I need -0.02v. It won't be long before I will need to remove the offset altogether it seems

Maybe thats the next step, wait for it to start failing without the offset and then RMA.

If its overheating at 0 offset this could well be still a sign its faulty and should be RMA'd if your using a 360 AIO (unless its mounted incorrectly)

 

[jigger]

That CPU will almost certainly fail prematurely as you seem to experiencing.
Running a 600 watt graphics card with these CPUs isn’t a great idea and has probably hastened the chips demise. IIRC Intel has some tools to test if it’s RMA time.

 

[no_1_dave]

That CPU will almost certainly fail prematurely as you seem to experiencing.
Running a 600 watt graphics card with these CPUs isn’t a great idea and has probably hastened the chips demise. IIRC Intel has some tools to test if it’s RMA time.

The problem is that Intel will just advise to use the default settings, which in their new bios limits the wattage to way below what was offered when new. Using their new revised and recommended bios settings the CPU is cut right down to just 34000 R23 score from the 40000 score it was when new. The updates they have released to 'fix stability' pretty much changes the CPU from a high end CPU to a mid-tier CPU. I think ED209 is right. I will need to wait until it fails with no offset, then complain to Intel and refuse to go along with heavily reducing the CPU performance because that's not what I paid for.

On a side note, I don't understand how the GPU being used would affect CPU longevity?

 

[jigger]

The problem is that Intel will just advise to use the default settings, which in their new bios limits the wattage to way below what was offered when new. Using their new revised and recommended bios settings the CPU is cut right down to just 34000 R23 score from the 40000 score it was when new. The updates they have released to 'fix stability' pretty much changes the CPU from a high end CPU to a mid-tier CPU. I think ED209 is right. I will need to wait until it fails with no offset, then complain to Intel and refuse to go along with heavily reducing the CPU performance because that's not what I paid for.

On a side note, I don't understand how the GPU being used would affect CPU longevity?

I’d try the Intel testing tools. If the chip fails Intel should accept the return.

A 600 watt graphics card and 300 watt CPU is essentially a two bar heater in box. This kind of heat load is a lot to ask of an ATX case.

 

[no_1_dave]

A 600 watt graphics card and 300 watt CPU is essentially a two bar heater in box. This kind of heat load is a lot to ask of an ATX case.

That's a fair shout but it should be ok as they're rarely both stressed at the same time...

When gaming the CPU rarely goes above 120w and most of the time it's lower than this and hovering around 20% usage
When encoding, the GPU is idle

The GPU is also undervolted to 450w

 

[Subliminal Aura]

On a side note, I don't understand how the GPU being used would affect CPU longevity?

Could it be more down to the stress placed through the motherboard and it no longer being able to stably deliver power ?
Think of it like a vinyl record getting warped over the years

Personally can't see it being the fault of the CPU

 

[Tetras]

So, is the clock ticking on my CPU?
Is there anything I can do about it without nerfing my performance?

So far as I know, the list of symptoms of a degrading CPU (from the published issues, I mean) aren't normally related to it needing a slow increase in voltage (or if it does, there are other issues alongside, like XMP/PCIE, crashes with nvidia drivers, crashes in Chrome), but like you said, best bet is to just let it degrade to the point it can't run stock anymore and then use the 5 year extended warranty to RMA.

 

[no_1_dave]

Could it be more down to the stress placed through the motherboard and it no longer being able to stably deliver power ?
Think of it like a vinyl record getting warped over the years

Personally can't see it being the fault of the CPU

I've changed motherboard since getting it when I moved from mATX to ATX

 

[Subliminal Aura]

I've changed motherboard since getting it when I moved from mATX to ATX

and you're still witnessing linear degradation over time ?

 

[no_1_dave]

and you're still witnessing linear degradation over time ?

Yep.
We will see how long -0.02v lasts. It would be nice if it stays stable, but something is telling me that within a month I'll be having to change it again

 

[ED209]

Yep.
We will see how long -0.02v lasts. It would be nice if it stays stable, but something is telling me that within a month I'll be having to change it again

To be honest, if it was me I would want it to fail as there is currently something wrong with it and for the amount you paid for it and their issues a nice new shiny 1 would be preferred!

You would not want it hobbling along and then killing over out side of warranty

 

[no_1_dave]

To be honest, if it was me I would want it to fail as there is currently something wrong with it and for the amount you paid for it and their issues a nice new shiny 1 would be preferred!

You would not want it hobbling along and then killing over out side of warranty

That's a good point

 

[jigger]

Personally I’d give the thing all the encouragement I could. These chips are cursed to fail over time, it’s just a matter of when.

RMA>14700k>sell platform/end

 

[Vince]

Surely the problem is also that anything they offer on an RMA 'will' go the same way, bit of a catch 22 really. I kinda agree with @jigger it might be time to either go 285k (which I think would be a mad move) or RMA it and consider your options.

 

[no_1_dave]

Personally I’d give the thing all the encouragement I could. These chips are cursed to fail over time, it’s just a matter of when.

RMA>14700k>sell platform/end

Surely the problem is also that anything they offer on an RMA 'will' go the same way, bit of a catch 22 really. I kinda agree with @jigger it might be time to either go 285k (which I think would be a mad move) or RMA it and consider your options.

I don't think selling the platform is needed, not yet anyway. There is nothing out there that makes it worth while.
Gaming at 4K and doing encoding is within just a few percent of the latest X3D chips and the 285 has worse gaming performance in a lot of games

With the warranty being 5 years I'll just see how it goes for now and RMA it if it stops being stable at stock. I hadn't realized they had extended the warranty.
Hopefully something worth upgrading to comes out before this happens so I can RMA then sell as new.