Overclocking and Life Expectancy?

[Hunched]

This may seem like a dumb question, but are there actually any real sources/information on how badly running your CPU or GPU at a 24/7 maximum voltage reduces lifespan?

Reason I ask is because everywhere I look, you can find tons of seemingly knowledgeable people saying opposite things.

 

There are people saying only temperature reduces, only voltage reduces, both reduce.

People saying it lowers lifespan by maybe 5% if that and others throwing out far larger numbers.

I've even seen some people say it is none of that, and it's actually temperature changes, going from 90c and back to ambient temps constantly, suggesting it's best to keep the machine always running like a server, keep a more constant average temperature.

 

Never sources, just people posting what they believe is common sense.

 

Though there have been people claiming they've mined with AMD cards 24/7 for 2 years straight at 90c-100c before having issues, which is reassuring for me as I'm doing nothing near as extreme.

 

 

 

You might wonder why I would even want to know this, why I wouldn't just simply not have my CPU and GPU running at fixed voltages and frequencies.

For my 970, the answer is simple and well documented if you look. GPU Boost makes it impossible to overclock, cannot even add +10mhz to the Core Clock before experiencing issues in less demanding games.

I can add +150mhz to the Core Clock and be fine, as long as I use my 970 only for running benchmarks or playing games all as demanding as Crysis 3, but run something where I'm getting 200fps+ and it will crash.

GPU Boost causes the voltage to lower too far in respect to the frequency when the GPU isn't being stressed enough which causes it to stop working, it's really that stupid.

So the choice is to use K-Boost in Precision X to force 1.2560v 24/7, or use a modded BIOS so it only does 1.2560v for 3D mode, or don't overclock...

 

My CPU... I was simply having issues with C-States causing stability problems and thought screw it, if I'm doing it to my GPU may as well to my CPU.

It's another thing where people are divided, with a good amount actually preferring a constant voltage on their CPU with no power saving at all.

 

My 4690k is always at Vcore of 1.280v-1.296v and is 35c-40c idle

My 970 is always at 1.2560v and is also 35c-40c idle

Obviously 100% load temps remain unchanged, which are both around mid 60c maximum for what it's worth.

 

Both are within the "safe" limits on voltage for what most people claim is before massive degradation begins.

I'm aware obviously degradation will occur from what I'm doing, what I don't know is if it will be massive or barely a scratch on lifespan.

I've seen people say at stock, Intel CPU's should last 8-10 years. If 5% decrease from what I'm doing isn't a made up statistic, that's barely anything.

 

Should I be concerned?

I'm thinking I'm just going to need to find out for myself, see how long my CPU and GPU last like this (this is my first PC) and when I upgrade, adjust accordingly if I did suffer issues.

I'm not really expecting any perfect answers from this thread, everyone everywhere else has their own experiences and stories which seem to vary drastically, opinions and all that good stuff...

 

 

 

 

 

 

[ImBleu]

As long as you're sensible, it'll be okay. If you cool your CPU with Liquid nitrogen trying to get to 6Ghz, it'll die pretty quickly. If you're more modest (for the 4690k leave it around max 4.6-4.8Ghz) you should be fine. Just don't be stupid with it.

[Guest]

As long as you're sensible, it'll be okay. If you cool your CPU with Liquid nitrogen trying to get to 6Ghz, it'll die pretty quickly. If you're more modest (for the 4690k leave it around max 4.6-4.8Ghz) you should be fine. Just don't be stupid with it.

And 4 ghz with a custom cooler?

[ImBleu]

And 4 ghz with a custom cooler?

 

Correct me if I'm wrong but isn't 4GHz the base clockspeed from the factory? if so then it's alright. 

[Guest]

Correct me if I'm wrong but isn't 4GHz the base clockspeed from the factory? if so then it's alright.

Not my CPU 3570K

[runit3]

I would snip and quote but im on my phone, sry.

Why would you think that what you're doing will cause degradation? The 1.3v rule is called "safe" because it's nearly universally accepted as the "max" 24/7/365 voltage. You'll upgrade your CPU before it dies. I had my 980X slightly beyond "safe" for 5 years, no degradation. The only instances of degradation that I've heard of or experienced are running in the 1.4-1.5v range for extended periods, extended being a few days or weeks, and even then catastropic instances are rare.

I'm leaving my new 4690k at 1.35v. If anything happens ill be sure to post it But I'm not worried at all.

[Hunched]

I would snip and quote but im on my phone, sry.

Why would you think that what you're doing will cause degradation? The 1.3v rule is called "safe" because it's nearly universally accepted as the "max" 24/7/365 voltage. You'll upgrade your CPU before it dies. I had my 980X slightly beyond "safe" for 5 years, no degradation. The only instances of degradation that I've heard of or experienced are running in the 1.4-1.5v range for extended periods, extended being a few days or weeks, and even then catastropic instances are rare.

I'm leaving my new 4690k at 1.35v. If anything happens ill be sure to post it But I'm not worried at all.

 

Good to know, I just had a bit of trouble looking for exactly the information I was looking for, it seems more common that people do use power saving on their GPU and CPU so voltages drop while idle or under less load.

I really don't care about saving power, and temps are amazing, the only reason I'd use power saving settings is for lifespan, but I've had issues with stability using them so it's more convenient to not.

If they can last years without issue, then I'm happy, I think I'll probably upgrade my CPU and GPU every 3-5 years at the latest.

 

Is it wrong to assume that the GPU will deal with safe voltages pretty much the same? Are they apples to apples or can that comparison with safe voltages and lifespan not be made between the two?

I suppose my GPU will be fine for years too judging from people mining 24/7 for years, way more stress overall than I'll be doing.

 

 

[runit3]

Good to know, I just had a bit of trouble looking for exactly the information I was looking for, it seems more common that people do use power saving on their GPU and CPU so voltages drop while idle or under less load.

I really don't care about saving power, and temps are amazing, the only reason I'd use power saving settings is for lifespan, but I've had issues with stability using them so it's more convenient to not.

If they can last years without issue, then I'm happy, I think I'll probably upgrade my CPU and GPU every 3-5 years at the latest.

 

Is it wrong to assume that the GPU will deal with safe voltages pretty much the same? Are they apples to apples or can that comparison with safe voltages and lifespan not be made between the two?

I suppose my GPU will be fine for years too judging from people mining 24/7 for years, way more stress overall than I'll be doing.

Haswell is not Sandy/Ivy, the C-states and Intel Speed Stepping OC issues have largely been resolved. I haven't been able to find any concrete evidence for them causing instability, and most OCers choose to leave them enabled with Haswell as they have no effect other than saving some power.

 

Older gen GPU's had similar issues OCing as CPU's. However, with the GTX 900's and to a certain extent the 700's they're all voltage limited. The only way you can actually overvolt them to a dangerous level is with a BIOS flash and editing.

[Cheddle]

Ive never had an issue with c-states causing instability on either my 4.7ghz g3258 or my stock (3.8ghz) i5-4690.

Gpuboost 2.0 on the other hand is bad... the stock tables don't work well with overclocks that rely on extra core voltage. Modded bios is the way to go here.

I prefer dynamicly changing voltages as it keeps heat (noise) to a minimum and results in my idle pc drawing 60Watts and full load drawing 350Watts.

it would bother me if my pc was drawing full power and at full noise constantly.

I leave my gear running 24/7 and ALWAYS overclock everything I can. I have never had a component die on me.

I personally believe that overly high temps over time and/or heat cycling are responsible for component failure.

my logic:

Solder and silicon soften at high temps, vibrations from fans, speakers and external case movment provide "shock" to softened metals resulting in distortions to the small soldering joints on the pcb. If unlucky these will create a short circuit either over time or as they cool down

further, when these materials are heated and cooled they shrink and expand and further deform, causing similar micro fractures and short circuits.

Modern capacitors have amazing life spans and huge temperate ranges. Cheap capacitors failing can be catastrophic. Depending on a capacitors use in a circuit over clocking could reduce their lifespan but only marginally... a cheap psu or a dirty power feed is much worse for a capacitor.

The reality is that these things last a LONG time... I have some old servers. Deployed in 2007 and ran 24/7/365 for 7 years under almost full load 5 days a week. They are 1U rack servers so they run hot and loud and they work perfectly... never had one fail. I attribute this to the fact that they are always on.