Raptor Lake degradation. For those that didnt know or werent able to keep up.

[Dragon1-1]

2 hours ago, LucShep said:

But AMD is not exactly exempt of the "single-core boost high voltages" either, or problems in their own platforms, you know...

I do, and the X3D series does run hot, but it's manageable, especially since the X3Ds also have fixed clock speeds due to the needs of their L3 cache. I'm pretty sure even the latest Ryzens aren't quite on the edge the way Intels are, and when undervolted, I'm sure my 5800X3D will last me a while. When the rig is due for an upgrade, it'll be to another X3D Ryzen variant.

[kksnowbear]

42 minutes ago, Dragon1-1 said:

the X3Ds also have fixed clock speeds due to the needs of their L3 cache

That's not accurate.  I have a 7800X3D with a X670E board, and even if we exclude what the CPU and board do on their own, some boards allow additional control over CPU clock (within reasonable limits, as with *any* board/CPU).  I bought and use the board I do for this exact feature (among others).

Not true to say you can't overclock an X3D CPU.

Edited July 29, 2024 by kksnowbear

[Dragon1-1]

You can, but not nearly as much as you can a CPU without 3D cache, and it's generally not recommended. So, you won't get extreme margin pushing on multiple stages like you do with Intel chips. If AMD doesn't do anything stupid, they should not undergo undue degradation.

[kksnowbear]

Nothing "not recommended" about it.  (Not recommended by whom? You?)

Perfectly safe and stable with proper cooling.  Plenty of info online from *very* reputable, far more knowledgeable sources, saying it can be done. Actually the CPU runs better (less power and heat, once it's dialed in properly) and performs better.

Been doing it now for many months on my primary machine, no problems whatsoever - even as compared side-by-side with a like platform that doesn't support the feature and thus can't benefit from the same overclock.

It's still overclocking, and it's (still) inaccurate to say X3Ds have a 'fixed clock', which is what you said.

(PS Probably not really a good time to compare Intel chips to...well, anything lmao)  The 7800X3D is already a better gaming CPU anyway, and doesn't destroy itself after 6 months

Edited July 29, 2024 by kksnowbear

[BitMaster]

Is there any law suit against Intel in this matter ?

It just can't be that you buy a faulty product and are left alone w/o a proper fix.

 

Intel should step up and recall them all, nothing else accepted imho.

[kksnowbear]

43 minutes ago, BitMaster said:

Is there any law suit against Intel in this matter ?

It just can't be that you buy a faulty product and are left alone w/o a proper fix.

 

Intel should step up and recall them all, nothing else accepted imho.

Not yet, that I know of...but you can bet your a$$ there are lawyers out there as we speak, just drooling  

[LucShep]

16 hours ago, LucShep said:

Thanks for the heads up, didn't even remember to check up Buildzoid's channel.

Watching the video right now, and there is exacty what I was saying before (crazy single-core boosts + voltages), check at about 12.13 and on:
 

...see what I mean?

It's insane how this happens.
These 13th and 14th gen i9s (and i7s?) all seem to work at stock with well over what is considered normal, reaching and surpassing 1.5v(!!).
12th gen i9s and i7s were already starting to slowly degrade if you're were using over 1.4v on long term overclocks.

I just wonder if it has to do with Intel probably using their own motherboards or BIOS settings for stability and long-term reliability testing, and not used by any other motherboard manufacturers (which could explain why they seem dumbfounded, and would be tremendous incompetent but not out of possibility). 
 

Last week I was in a group chat where this stuff was being discussed, there was a video posted there about a degraded i9 13900KS.

Absolutely worth the watch, more so if you already watched Buildzoids' oscilloscope video (the one I'm quoting) where the crazy single-core boost voltages problems are shown.

If all you watch are the glorified techtubers in the techspace, this guy will look obnoxious (ignore that).
The point here is, he's actually correct and his aproach here makes all the sense. 

:WARNING: swearing

 

Edited July 30, 2024 by LucShep

[BitMaster]

I watched Der 8auer's video last night regarding this topic and it was eye opening.

1: Any CPU already showing symptoms cannot be fixed, period !  Intel's POV btw.

2: Market value of used 13th/14th CPU is null. No one knows how long a CPU will run before it has degraded too far.

 

They face a hard time...Intel needs to come with a solution for this dilemma. It's not done and good is like with Meltdown/Spectre, this is a different beast.

 

I am sooo happy I havent touched ANY new Intel since Ryzen emerged.

 

[LucShep]

31 minutes ago, BitMaster said:

I watched Der 8auer's video last night regarding this topic and it was eye opening.

1: Any CPU already showing symptoms cannot be fixed, period !  Intel's POV btw.

2: Market value of used 13th/14th CPU is null. No one knows how long a CPU will run before it has degraded too far.

 

hmmmm  ....I'm looking at it the other way.... I may start looking at (then) reeeeeally cheap 14900Ks, then mess around with BIOS settings sometime soon. 

[okopanja]

Any confirmed DCS world fatalities?

[LucShep]

1 hour ago, okopanja said:

Any confirmed DCS world fatalities?

Not sure how many will raise the arm to confirm, but......   

The truth is that this is really problematic, even more than expected. And, yes, there'll be fatalities among DCS users' Intel 13th/14th gen CPUs at some point, if not already.

If there's a techtuber that you should take seriously is Roman “der8auer” Hartung, an extreme overclocker that is also a mechatronics engineer.
He did what most should have done, which is taking his time and only release an opinion that he's sure on and ready to be released (contrary to most who ran for the clicks).

This is what he has to say (at 11.02 in the video):

It seems it affects all 13th and 14th gen CPUs that are above 65W. 
 

Edited July 30, 2024 by LucShep

[SharpeXB]

55 minutes ago, okopanja said:

Any confirmed DCS world fatalities?

It’s a good guess this is why my 13900K went kablooey. I noticed some really severe slowdown with it and just replaced it with the 14900KS. Out of the pan and into the fire That one shows no sign of trouble… yet. 

[LucShep]

2 minutes ago, SharpeXB said:

It’s a good guess this is why my 13900K went kablooey. I noticed some really severe slowdown with it and just replaced it with the 14900KS. Out of the pan and into the fire That one shows no sign of trouble… yet. 

Yes, your 14900KS will need the microcode update, that's for sure. Don't trust lady luck on this.

Everybody with 13th and 14th gen CPUs that are over 65W needs to do it once it's out (no buts or ifs).

Get the friggin microcode update installed once it's out. And consider RMA your CPU if there are signs of damage already.

[BitMaster]

The really problematic thing is that, as explained in 8auer's video, you cannot trust them ( the CPU, actually Intel too imho ) anymore, not further than from now to 1 second ahead, if at all.

And as I said before, I dont think DCS users will be helpless, we are among the few percent that usually know PC's, it's the womanwith 3 kids next door, the old man down the road, novice users.... Intel must reach out and take them by the hand. I fear they will only do this if forced by law and a judge and then only as far as they must to avoid penalty payments.

They knew it was their own fault and still blamed others.... Der 8auer left that part out. Better if he still wants to be able to phone Intel up and expect them to not hang up 1st second LoL

Any 13th or 14th CPU above 65w is doomed...and maybe more SKU's...  time will tell, what a freaking show from Intel  LoL

Edited July 30, 2024 by BitMaster

[Dragon1-1]

On 7/29/2024 at 6:02 PM, kksnowbear said:

Perfectly safe and stable with proper cooling.  Plenty of info online from *very* reputable, far more knowledgeable sources, saying it can be done. Actually the CPU runs better (less power and heat, once it's dialed in properly) and performs better.

You're describing undervolting, I do that, too. It helps with X3Ds' tendency to run hot. Naturally, better thermals will usually result in higher average clock speeds. Also, undervolting doesn't kill your CPU, quite the opposite, in fact, so the underlying point still stands.

I admit, it's not quite a fixed clock (which hasn't really been a thing since a long time ago), but they do have a locked multiplier. Either way, the "classic" overclocking done by combining an oversized cooler with an overvolted CPU and looking how far you can take the die is not really a thing with the X3Ds.

[kksnowbear]

 

5 minutes ago, Dragon1-1 said:

You're describing undervolting, I do that, too. It helps with X3Ds' tendency to run hot. Naturally, better thermals will usually result in higher average clock speeds. Also, undervolting doesn't kill your CPU, quite the opposite, in fact, so the underlying point still stands.

I admit, it's not quite a fixed clock (which hasn't really been a thing since a long time ago), but they do have a locked multiplier. Either way, the "classic" overclocking done by combining an oversized cooler with an overvolted CPU and looking how far you can take the die is not really a thing with the X3Ds.

Nope.

I'm talking about increasing clocks on a 7800X3D, same as overclocking has always been done. And it's not a fixed clock either way.

Nothing to do with undervolting, directly, nor locked multiplier. 

(lmao Save us both a lotta time if you just go ahead now and acknowledge you don't have any idea...)

Edited July 30, 2024 by kksnowbear

[LucShep]

4 hours ago, BitMaster said:

The really problematic thing is that, as explained in 8auer's video, you cannot trust them ( the CPU, actually Intel too imho ) anymore, not further than from now to 1 second ahead, if at all.

And as I said before, I dont think DCS users will be helpless, we are among the few percent that usually know PC's, it's the womanwith 3 kids next door, the old man down the road, novice users.... Intel must reach out and take them by the hand. I fear they will only do this if forced by law and a judge and then only as far as they must to avoid penalty payments.

They knew it was their own fault and still blamed others.... Der 8auer left that part out. Better if he still wants to be able to phone Intel up and expect them to not hang up 1st second LoL

Any 13th or 14th CPU above 65w is doomed...and maybe more SKU's...  time will tell, what a freaking show from Intel  LoL

 

Absolutely. It's a total <profanity> show. 

This stuff can be pretty aggravating, as many users kept working from home since the pandemic, and are oblivious to this crap. Not funny if your daily working tool uses one of these Intel chips and goes "kaput".   

I'm not so sure about DCS users, how many visit this section anyway?  There may be so many who may also be oblivious to this.

At least some kind of heads-up to friends, then them to their friends and etc, also in other communities, should be done by each of us. So many people who should be alerted.

Meanwhile, another section of the freak show has already started... all of the 12th gen "K" CPUs have already started to slowly go up in price (at least where I'm looking). 

Edited July 31, 2024 by LucShep

[Dragon1-1]

9 hours ago, kksnowbear said:

(lmao Save us both a lotta time if you just go ahead now and acknowledge you don't have any idea...)

Only after you tell me how you are overvolting your CPU and getting better temps that on stock settings. So yeah, lmao. If you're trying to tell me you're cheating thermodynamics, pull the other one. Next you'll tell me it runs on cold fusion for cheap power.

You're undervolting the CPU. That's the only way anything you said makes sense. You may also be increasing the boost clock manually (note that this is not something AMD says you should be able to do), which can let you squeeze it for a bit more, but it's not on the same level as what you can do with a multiplier-unlocked CPU. I don't bother with that because the increase is marginal in real applications. And I can't help but notice that all this time, you failed to address my core claim: any undervolted configuration is going to be less demanding on CPU when it comes to wear than defaults, even if boost clock is increased.

It's simply not the kind of overclocking we're talking about when it comes to Intels. People most affected by the degradation crank up voltage and use it to run up the multiplier, giving a considerable increase in clock speed, at expense of power consumption and heat. Since Intel had the bright idea to do more or less that out of the box already, it's easy to reach the point at which silicon simply can't keep up for very long. X3D series Ryzens do not have this problem, as the tuning they usually get is undervolting.

[kksnowbear]

2 hours ago, Dragon1-1 said:

Only after you tell me how you are overvolting your CPU and getting better temps that on stock settings. So yeah, lmao. If you're trying to tell me you're cheating thermodynamics, pull the other one. Next you'll tell me it runs on cold fusion for cheap power.

You're undervolting the CPU. That's the only way anything you said makes sense. You may also be increasing the boost clock manually (note that this is not something AMD says you should be able to do), which can let you squeeze it for a bit more, but it's not on the same level as what you can do with a multiplier-unlocked CPU. I don't bother with that because the increase is marginal in real applications. And I can't help but notice that all this time, you failed to address my core claim: any undervolted configuration is going to be less demanding on CPU when it comes to wear than defaults, even if boost clock is increased.

It's simply not the kind of overclocking we're talking about when it comes to Intels. People most affected by the degradation crank up voltage and use it to run up the multiplier, giving a considerable increase in clock speed, at expense of power consumption and heat. Since Intel had the bright idea to do more or less that out of the box already, it's easy to reach the point at which silicon simply can't keep up for very long. X3D series Ryzens do not have this problem, as the tuning they usually get is undervolting.

Absolutely none of which is the point.

You said X3D CPUs have a fixed clock, and that's not accurate.  I didn't say anything about whether it's the same as what people do with these Intels.  I didn't say any thing about levels it can be done with unlocked multipliers, nor whether you do or don't.

All that is just plain old distraction.

I said your statement about X3D CPUs having a fixed clock is inaccurate.  And it is.

You said what I mentioned was "Not recommended" ...but never indicated by whom.  You now say it's "Not something AMD says you should be able to do".  Where do they say that, exactly?  By all means, enlighten us all.

See, what you're doing there is trying to sound authoritative...but I doubt there's any real accuracy in these comments.  I could be wrong, but I doubt it.

LMAO And after all that, it's pretty obvious that you have no idea.  Got it.

Edited July 31, 2024 by kksnowbear

[Dragon1-1]

1 hour ago, kksnowbear said:

You said X3D CPUs have a fixed clock, and that's not accurate. 

And in the very next post I corrected myself, which you would have known if you bothered reading and comprehending what I said.

1 hour ago, kksnowbear said:

Absolutely none of which is the point.

I stated the point multiple times. To reiterate, it is that due to fixed multiplier and general lack of OC options, Ryzen X3D series is not susceptible to similar degradation to Intel. Yes, that is the point. You're fixating on a minor misuse of a single term, because that's the only thing that you did get right, while choosing to ignore the real point even after I served it to you on a platter.

I have one of those CPUs myself, and yes, I did tune it. I've been doing this for a while, too, albeit on Intels, because everyone used Intels before Ryzen came along. If you think you have some mystical OC knowledge that I do not... well, you really don't. In the context of how Intel versus AMD are at risk of hardware wear, undervolting is not a relevant OC technique. That is the point, and it is so because any undervolting decreases the wear as opposed to increasing it. Please address the bolded part, instead of latching onto a minor terminology slip.

[kksnowbear]

So you cannot provide any authoritative reference for your own statement that it's "not recommended".

Likewise, you cannot provide any reference to corroborate your statement that what I'm doing is "Not something AMD says you should be able to do".

And you still clearly don't know what I'm talking about.  As a hint, you can't do it on a 5800X3D (to my knowledge)...so you're now saying you actually have done it on the 7800X3D, which you're saying you also have?

PS I read and comprehend just fine, thanks.   What I don't do is fall for it when someone pees on my leg and tells me it's raining:  If you had just corrected yourself, and if it were just a "minor terminology slip" as you now say, then it wouldn't be necessary to say in the very same post that it's "not recommended". See? You didn't just correct yourself for an isolated slip.  You 'doubled down' on yet another inaccuracy, just to make it sound as if you were right in the first place.  And then you furthered it with your statement about what AMD said.  It's not really just correcting yourself when you throw more inaccurate assertions on top.

But let's assume I'm wrong...by all means, tell us: Not recommended by whom, exactly?  And where does AMD say what you indicated?

Edited August 12, 2024 by kksnowbear