Hot CPUs and longevity

[P4-630]

Warranty is 2, right?

Intel, 3 years.

 

[Deleted member 185158]

The example

If FIT = 15.1, then MTBF = 10^9/15.1 = 66225165

Is roughly 18 years. The end result is seconds of time FTW.

Degradation is only an overclockers term. The values given is with stock default hardware.

The caps on your motherboard is likely to go bad or "die" before your processor does. Whether or not the board takes out the cpu is a different story/topic.

 

[CraptacularOne]

Every paradigm shift people freak out because what they are seeing isn't what they are used to. This is no different and actually the way they have been designed for years now. They've always been rated to operate at 100C. They forget that most laptop CPUs run near pegged at 100C for their entire life because of cooling constraints and have no longevity problems what so ever. In fact it's the battery in laptops that gives up way before the silicon in the CPU does. Replace the battery and it'll be good to go.

I know lots of people with desktop systems in service for 10+ years on the stock cooler or some random aftermarket garbage cooler that do things like video editing/rendering, code compiling and other heavy tasks on the regular and their systems are fine despite the CPU being pegged at nearly 100C. Most I've done for them is a RAM upgrade here or a storage upgrade there.

 

[Deleted member 185158]

According to an article at a competitive site.....

• Ryzen 5000 series fails at 2.9 percent.
• Ryzen 3000 series fails at 3 percent.
• ThreadRipper 3000 series fails at 2.5 percent.

For comparison, the company’ data on Intel chips:

• Intel 9th-gen fails at 0.9 percent.
• Intel 10th-gen fails at 1.2 percent.

Ryzen 5000 failure rates: We reality-check the claims

One system vendor reported that Ryzen 5000 chips are failing at a high rate. Our sources and others suggest the problem is isolated.

www.pcworld.com

 

[Vayra86]

The example

If FIT = 15.1, then MTBF = 10^9/15.1 = 66225165

Is roughly 18 years. The end result is seconds of time FTW.

Degradation is only an overclockers term. The values given is with stock default hardware.

The caps on your motherboard is likely to go bad or "die" before your processor does. Whether or not the board takes out the cpu is a different story/topic.

Boards can come with stock settings pushing near or over 1.4V through a chip and readily bump into thermal limits now... and chips are practically OCd out of the box...

 

[Deleted member 185158]

Boards can come with stock settings pushing near or over 1.4V through a chip and readily bump into thermal limits now... and chips are practically OCd out of the box...

This means nothing by the design.

It's obvious the cpu can withstand the melting point of solder. IE: IHS plates....

 

[dgianstefani]

Voltage kills CPUs, not temperature (CPU will shut the system down long before temperature becomes an issue that could cause damage).

One relevant thing to bear in mind if you're chasing an OC is that lower temperatures reduce voltage leakage, so the cooler you run your chip, the less voltage you need to feed it to achieve the same frequency and stability (although this is a relatively minor difference).

The main temperature sensitive parts of any system are the power delivery (PSU and VRM) and RAM/VRAM/SSD storage. These should be kept cool if possible. Ideal RAM temperature is below 35 °C, PSU/VRM the cooler the better up to a point, around 20-30 °C is "ideal" but hard to achieve. SSD storage should be kept below 75C if possible, NAND flash actually performs best above a certain temperature, but higher temps can degrade. AFAIK RAM stability is tied to temperature, B die in particular gets unstable above 40 °C.

 

[Shrek]

High temperature leads to diffusion so can be a long-term issue, but voltage is a very important factor.

 

[Deleted member 185158]

Voltage kills CPUs, not temperature (CPU will shut the system down long before temperature becomes an issue that could cause damage).

One relevant thing to bear in mind if you're chasing an OC is that lower temperatures reduce voltage leakage, so the cooler you run your chip, the less voltage you need to feed it to achieve the same frequency and stability (although this is a relatively minor difference).

Accurate statements.
The convo is about temps, not voltage (wattage).

CPU isn't damaged by such low temps.
Because I delid all my chips via soldering torch
My overclocking degrades them.
That's FX chips and 6/7ghz club overclocking ftw...

 

[Vayra86]

Absolutely correct. And what have we seen with people who see 100C or other peak temps, they improve cooling. And with improved cooling more voltage is possible... Ive seen so many recent examples of people pushing chips over 1.4v for 24/7... theyre going to see it degrade.

 

[Shrek]

Diffusion is a long-term process, so not so much of an issue for short-term soldering temperatures.

 

[Deleted member 185158]

Absolutely correct. And what have we seen with people who see 100C or other peak temps, they improve cooling. And with improved cooling more voltage is possible...

Running the cpu at lower temps is very desired. But not necessary.

Most of us overclock. So a lot more mindful of our actions and cooling preferences.

I like 2.0v and LN2.

 

[freeagent]

I have seen it on my 3570K. Near 2018 it could no longer do 4.4 Ghz under the set voltages. Wanted more vcore for stability but I reduced clocks to 4.2 instead at much lower volts. And that 4.4 was with 86C peak temp

I only know this because I got my 3770K around that time. They were pushing out mad mitigation updates for specter and meltdown, starting in late 2017. I watched them shave GFlops from my 3770K. In the end it required more voltage for X clock, and I lost 100MHz off the top. Microcode was also updated, and it pushed more vcore after that update.

 

[Vayra86]

I only know this because I got my 3770K around that time. They were pushing out mad mitigation updates for specter and meltdown, starting in late 2017. I watched them shave GFlops from my 3770K. In the end it required more voltage for X clock, and I lost 100MHz off the top. Microcode was also updated, and it pushed more vcore after that update.

Dunno I never did any mitigation updates nor bios... and max delay on Wupdate. But interesting nonetheless!

 

[freeagent]

Dunno I never did any mitigation updates nor bios... and max delay on Wupdate

You didn't have to, it was pushed through windows update

 

[mclaren85]

I also worry about my Ryzen 3900x chip. So I disabled the Core Performance boost and set the Windows Power Settings to "Power Saver" in order to prolong the life of the CPU. Now, my cpu runs at 1700 MHZ and 38C Celcius

 

[Deleted member 185158]

I also worry about my Ryzen 3900x chip. So I disabled the Core Performance boost and set the Windows Power Settings to "Power Saver" in order to prolong the life of the CPU. Now, my cpu runs at 1700 MHZ and 38C Celcius

View attachment 266645View attachment 266646

I've done the same on my son's 2700x.

Or alternatively, because he doesn't really need all 16 threads, I shut down 4 cores and run the cpu at defaults with boost and nearly cut the wattage in half. (Food for thought).

 

[The red spirit]

Degradation will happen eventually at peak temp. And if you then add voltage to counteract that, you will slowly degrade it more. Think 5-7 years depending on node and voltages used. Which just so happens to generally be the age at which youll want something new.
Its not a major issue, but yes you can find it and high temp/voltage will decrease longevity. Whether you will be bothered is personal/use case.

It's basically slow electromigration, but also let's not forget many hot-cold cycles and the fact that at higher temperatures chemistry gets more reactive generally and CPUs have tons of different elements.
You can watch this vid if you want it goes a bit deeper into heat problems:

Regarding what we actually use, some Zen + or Zen 2 chips actually had electromigration problems and early failures, basically Pentium 4 sudden death style. AMD did a quiet recall and was trying to cover it up hard, saying that it wasn't electromigration, but some boost bug. Some experiments with way too high voltage were done by some people and yeah, it was possible to kill those chips very fast, despite that, at normal boost conditions, some chips croaked before hitting two years or even a year. Like I said, it was rectified quickly, but it certainly shows how basically we are on knife's edge in terms of durability. Also some overclockers noticed that Sandy Bridge chips at 5GHz tend to degrade really fast and after a year or two fail to function at stock settings.

I also worry about my Ryzen 3900x chip. So I disabled the Core Performance boost and set the Windows Power Settings to "Power Saver" in order to prolong the life of the CPU. Now, my cpu runs at 1700 MHZ and 38C Celcius

View attachment 266645View attachment 266646

It doesn't run at 1.7 Ghz, it's just idle state, but if you seriously locked it at that speed, then you are too paranoid about it.

 

[TheoneandonlyMrK]

You didn't have to, it was pushed through windows update

They very kindly updated my 8750H laptop to a point where I am questioning a update of hardware, it has noticeably slowed down.

 

[freeagent]

They very kindly updated my 8750H laptop to a point where I am questioning an update of hardware, it has noticeably slowed down.

I know, it’s a bit sad.. I know how you feel.

 

[mclaren85]

It doesn't run at 1.7 Ghz, it's just idle state, but if you seriously locked it at that speed, then you are too paranoid about it.

Correct! That is the idle speed. I do sometimes see a full 3,7 Ghz when processing

 

[The red spirit]

Correct! That is the idle speed. I do sometimes see a full 3,7 Ghz when processing

That shouldn't be sometimes.

 

[mclaren85]

That shouldn't be sometimes.

I can't always afford to buy a Ryzen 9 3900x or equivalent Cpu

 

[The red spirit]

I can't always afford to buy a Ryzen 9 3900x or equivalent Cpu

It won't die THAT fast.

 

[caroline!]

My old Sempron idled at 66C when overclocked, there was apparently a sensor bug in the mobo too because it wouldn't read numbers higher than 70, it was probably running at 100 degrees all the time when I was running a game, it was like that for about 5 years, it still works.