Do differences in "safe level" GPU temperatures affect lifespan?

  • Thread starter Thread starter Guest
  • Start date Start date
G

Guest

Guest
So, when it comes to cooling, having a silent system is always nice but if you ask me temperatures are always the first priority. 80s and 90s (Celsius) are regarded as "safe level temps" but if you keep the GPU cooler than that it'll likely last longer. What I want to know is: will there be a difference in lifespan of cards that under load sit at 70, 60 or let's say a watercooled 50 degrees? When you get into that safe zone do differences still matter in terms of longevity? I'm currently choosing a case and air cooling the GPU is my No 1 priority so I'm kind of having a tough choice...
Thanks for any information shared!
 
Solution

Manufacturing and handling defects aside, semiconductors don't die unless their safe operating area is exceeded and semiconductors won't exceed their SOA unless their specifications are ignored (overclocking, overvolting, poor design, etc.) or something else fails and cause the specifications/SOA to be exceeded.

The only difference between 80C and 100C is that you have that much less error margin in case something goes wrong. For example: drivers crash and leave the GPU in a...
The stock coolers are designed to be safe for years. So my question becomes does it matter if lifespan is affected given that they already last more than a few years? Do you really want to be running a 4-6 year old GPU just because it hasn't died from heat exhaustion?

I have a Radeon 4890 that still works just fine. I don't use it because it's 2016.
 


I'm not saying you should be using a 4+ year old GPU :), I just wanted to know if there was a temp above which the degradation starts to be noticeable. Let's say 50 and 70 degrees difference...
 
You probably have a higher probability of killing your GPU in your pursuit for lower temperatures "for longevity's sake" from liquid cooling loop failure than overheating with the stock HSF.

Unless you decide to overvolt and overclock it, the stock HSF and settings should be perfectly fine and the GPU will easily outlast its useful life. Plenty of GPUs have survived years in bitcoin/altcoin miners running at 80-90C and that would be 24/7 heavy load, not intermittent gaming.
 


So you think it wouldn't be worth it buying a case for getting around 10 degrees cooler gpu on air? I'm in a tough choice, basically I wanted a Define R5 case and when I saw the Enthoo Pro (which is even cheaper) runs the GPU 11 degrees cooler than R5 I started wondering... Arc Midi R2 is somewhat better solution than the R5 but I still couldn't decide. I really like the Enthoo Pro but I think I'd be better off with a smaller case. What do you think?
 
General consensus has it that Intel cpu's have a soft cap safe limit of @70°C and gpu soft cap limit of 80°C. This is as high as you should go for any extended period or you start running into the risk of the silicon degrading at a much higher rate. So yes, this does affect lifespan. I own a pentium II-350MHz that's over 18 years old, is overclocked to 400MHz and still works fine. What then in a cpus lifespan. It's generally nothing physical. It's everything technological. With just 4Gb of ram, Windows 98SE, an AGP x8 graphics slot, its so outdated it's ridiculous. Zero chance of any upgrade. So old it won't support a 1Tb hdd.
So think 5 years is about good. Keep the cpu under 70°C and no worries.

A word on stock cooling. Stock coolers are designed for 1 thing. Keeping the cpu under @70°C under nominal usage. That means office work, Ms hearts, Web surfing etc. Not rendering videos. While a cpu might be capable of such, the cooler isn't. Anything over @70% cpu usage is considered extreme, as in beyond nominal, so seeing temps on a stock cooler climb past 100°C is to be expected.

The cpu doesn't care what it's temp is, as long as it's under @70°C. 50° is the same as 30° or anything in between.

What does affect cooling ability is ambient temps. In an air conditioned room at 22°C, its normal for cpus to idle @30-40°. If the ambient temp is already close to 40°, you can expect the pc to idle @50-60°C. So cooling is relative to needs.

Liquid or air, doesn't matter, does the same job at stock speeds, but size does. The hotter the cpu, the bigger the cooler. Sometimes it's just not feasible to run on a stock cooler. Even at stock speeds.
 


Thanks for the reply! I see you own the R5, so could you please give a few words about GPU cooling in it?
 
My general rule of thumb is to get the quietest case and HSF setup for the cooling. I agree with keeping the temps under 80 on the GPU and under 70 on Intel CPU's. After that point, if you have a super cool setup, you can make it quieter with less fan speed, or OC higher.
 
My rig. I7 3770k OC 4.6GHz on nzxt Kraken x61 liquid cooler. Asus gtx970 OC to max levels. Fractal Define R5 window. Sits in a cubby under my desk. Idle temp is @32-34°C cpu, 34-36°C gpu. Prime95 100% load on cpu is 63°C. Hardest gpu hitting game I own puts gpu to 99% usage, 72°C.

There isn't an issue with cooling on the R5. The only issue is in those tests are run with all fans on maximum and overloading the heat. My cpu fans never go over 700rpm, they max at 1000. My gpu fans are custom curved, never get above 70%, don't even turn on until 65°C. So is that 11° an issue? Absolutely not. It's only an issue if you feel inclined to sit and listen to every fan at max, which it never gets to.

Edit. It's so quiet at idle, the only way I know the pc is actually on is from the power/hdd light. Can't hear it otherwise. Even running a game, its almost inaudible. Unless I push it under stress, I barely hear it.
 


Thanks for the great description! :) Do you have a bottom intake? I think a bottom fan improves GPU temp a lot so it's the first fan I'll get. Other than that I believe I'll be fine with an R5 if I get a card with a great cooler (my eyes are currently on the 1070 Xtreme Gaming). Also, are the noise levels reasonably low even under full load? After all that's the purpose of this case...
 
Ok. Here's my fans. 2x stock fans at intake, governed by MSI software from my motherboard. Cap on them at 100% duty cycle is 800rpm, low cap is 450rpm. 2x stock 140mm nzxt fans from the kraken x61 on top exhaust. Governed by Cam software through the cooler. That's it. No rear exhaust, no bottom exhaust, just 2x 140mm intakes and 2x 140mm exhaust.

Now because I use the radiator top Mount, I do loose the baffling that's stock there. It's trixy. Understand, a rear exhaust does nothing really to help top mounted fans. Any heat given from the gpu travels UP the side of the case, then out the top, cpu heat, out the top, case heat, out the top. Trying to force the exhaust 90° and out the rear is honestly pointless. Heat rises, gpu blows it up, fans blow it out. Simple. Natural progression. For me.

IF you decide on an air cooler (I'd suggest a cryorig R1, Phanteks ph-tc14PE or Noctua NH-D14), that changes everything. Those coolers draw huge amounts of air and direct it right at the rear exhaust. So. You could leave all the baffling on the top, run a single 140mm exhaust at elevated rpm and 2x 140mm intakes at reduced rpm. Vast majority of any gpu, case heat will naturally be directed to the rear, which won't be that loud realistically, the double intakes can spin slower and still provide ample cfm and the case is completely baffled. In this, it would be quieter than mine at full load since my exhaust is on top, so sound is right there vrs exhaust out back where sound is facing away.

Any of those big twin towers are capable of cooling @220w or better of thermal heat. Your cpu will be 100w or less, give or take, unless you OC. At stock or close to stock speeds, the overwhelming ability to transfer and dissipate heat of those coolers will mean they barely get warm. So their fans don't spin fast. Silent. Stock coolers are small, have barely adequate ability to transfer and dissipate heat. Get very hot. Fans spin really fast. Tons of noise. Oversized gpu is the same theory. If it's barely working, its not getting hot, so the fans (if you even need to turn them on) make no noise.

The only thing in a pc that really makes any noise is the fans. Find a way to slow them down, they get quieter. Anything under @900rpm is relatively quiet on average. Under @600rpm is inaudible. Pretty much.
 


Well, HDDs make no noise but I assume you removed the cages although I might use one and the fact they're designed so good not to interrupt airflow is a plus. I think intake at the bottom would help GPU because it's fresh air they're pushing in, but that could also compromise the silence a little bit.

About CPU, I'm definitely getting an air cooler, 212x seems just fine and you can't get a liquid one for that price. Plus, my CPU really won't need extreme cooling, that's for sure. :) I'll probably use at least one vent at the top for exhaust, if it adds a bit of noise I'm fine. Now, I really don't know whether I should use the motherboard headers or manual control or something else for the fans. Also, I don't know how I can control the bottom one based on GPU temperature because very few cards have fan headers on them (I've only seen the Asus one).
 
I'd. Other use a bottom mount fan, just the 2x front intakes. The bottom mount is towards the front of the case, so what end up happening is when that fan sees higher rpm, its going to push a good chunk of the front intake air 90° up, wasting it. The upper of the 2x intakes is right in front of the gpu, and yes, I did remove one of the hdd bays (the taller one)
As for cooling via gpu, yes, that can be done by any card. There is a program called SpeedFan. It's a nuisance to originally set up, definitely takes some getting used to, but has a single advantage over any other design. It works off the address of the sensor, not the sensor itself. Every sensor has a physical address, the hard part is figuring out exactly what address goes to what header. So you can use the gpu temp address, and redirect it to any header. It also has the ability to make custom fan curves, at many different points, so setting fan speeds according to close temp ranges is easy. So you can easily set Sys_fan 1 header to respond to gpu, or motherboard or even cpu temps or change the cpu_fan to respond to case temps. It's quite adaptable.

Honestly, I like the performance of the hyper212. Vrs it's price point, but I much prefer the Cryorig H7. It's shorter to the board, offset so it sits next to the ram, not above it, outperforms the 212 in every test and is quieter. All that's worth a few bucks more to me.

Liquid coolers don't have to be extreme. My wife's pc is a i5-3570k at 4.3GHz. It sits in a CM 690 II Adv case that's really nothing more than a frame covered with grill. There's absolutely no sound control. It's cooled by a corsair h55 (thin 120mm radiator, exhausts out the top) with a Noctua fan. Can't hear it. At all. Ever. If I push the system, the fans on the very quiet Asus gtx 660ti are loud enough to drown any cpu fan noise easily. My wife can put her phone on speakerphone, stick it on top of the pc and she'd get more complaints from the almost none noise of the soft key keyboard than she would from the pc itself.

Silent pc's are very easy to make, all you need is plenty of airflow through the case (use as many and the largest fans possible) , use slow fan speeds and over-cool the cpu. Just don't screw up the airflow, air needs to go in and out with minimal obstruction or you defeat the purpose of the fans and get a circulatory airflow, not a linear one.
 


I really wanted the H7 but it's not sold in my country. A bad thing that just came to my mind is how short does the PSU have to be so you can install the middle bottom fan? On this photo it looks like it should be very short and that's bad cause I'll get a G2 550W... 🙁 (http://www.pdconsec.net/Data/Sites/1/media/blogs/davidr/define-r5/define-r5-review-11.jpg). The CS750M on the photo is 140mm long, so will a 165mm PSU and a fan fit? I thought about a side fan but I've heard it interrupts the fans of the graphics card because it's side exhaust (on aftermarket models).

Also, if you look at the photo above, it looks like the bottom fan blows directly into the GPU so it doesn't waste any front intake air.
 

ATX PSUs are all the same dimensions except for length and the 170mm length is something you only need to worry about if you plan to use a 1200-2000W PSU. How big the PSU fan might be is irrelevant since it is inside the PSU.
 


I wasn't talking about PSU fan but the bottom fan in the R5. It says you can only install it if the PSU's under 170 long.
 
I have a 550w G2 in my R5 lol, great psu. There's no issues with the fan there. It'll physically fit, you just have to make allowances for the modular plug wires, wiring does not like to bend sharply at 90° right out of the connector, so sits right above the edge of the fan. Not an issue if the wires are kept neat and tidy, but could be an issue if just slapped in there. Same would be true of any ATX psu.
 


Phew, that's a relief. 😀
Yeah, I don't think there's a better PSU for anyone who's fine with 550W. Tier 1 quality for that price... thumbs up! By the way, do you use it fan up or down? Which will be better for GPU temps?
 
Down. You want the heat to rise. Sticking the fan up, you'll be using the psu draw to create a vacuum under the gpu. If you get that fan spinning fast enough, (like under heavy loads) you'll start impacting the gpus draw and exhaust, possibly to the point where the up draft equals the down draft and all that gpu heat will basically sit right there at the gpu. Fan up is only used in dire circumstances or if the case has no external psu air vent. Best to keep cool low air blowing through the psu, untainted by excess heat from inside the case. Psu's are tested at 25-50°C heat. No sense adding 70°+ C gpu exhaust to the mix.
 

Even in cases with no intake dedicated to the PSU (ex.: Antec 300v1), there is 1-2cm of clearance below the PSU for the PSU to draw air from within the case. Unless the PSU has horrific efficiency, that's more than enough clearance along three sides to draw enough air to keep the PSU reasonably cool with minimal PSU airflow resistance.
 
Very generally speaking re: all electronics pretty much....heat is enemy #1 and it doesn't have to be excessive heat necessarily. 'Moderate' heat over time can take its toll.

Example:

Just pulled the trigger on a new Sapphire RX 470 gpu. It's actually going to retire a 4850 I've been running since 2008...that's 8 years. It started life as a stock cooled Asus HD4850 for $200. It had that gawd awful single slot 'thin' blower style cooler that sounded like a blow dryer/leer jet when it spun up...loudest cooler I've ever heard to this day. As soon as I heard it I thought 'oh hell no' and ended up ordering an Arctic Cooling Accelero Twin Turbo HSF. (A rock-solid cooler that lasts btw....lol)

IIRC...Two 92mm fans on top of a 4 pipe cooler and it came with an assortment of alum stick-on sinks for the vram. Everything pretty much fit like it should...sorta. It did leave the vrms uncovered. Had some old low profile copper 'slug' sinks that I filed to fit and used epoxy type thermal compound to perm. mount. It was/is a one size fits all cooler so it was a lil (lot) overkill for a lil' 6pin 4850...but as a result, I believe, it lasted so long. The rig it was in ran 24/7/365. First an E8400 C2D/P45 @ 4.24GHz...then my 2500K/Z68 @ 4.6GHz. I truly believe being 'over-cooled' had a lot to do with a mid-range card lasting 8+ years running 24/7/365...general use and moderate gaming during the day and seeding torrents all night. It's now retiring to 'spare' status in perfect working condition.
 

Semiconductors will last almost indefinitely under 100C junction temperature. When electronics fail, it is usually due to something else failing first, that other thing usually being capacitors either on the card/motherboard itself or the power supply that feeds it.

My HD5770 has probably seen something close to the equivalent of three years of 24/7 full load (much of it from sitting inside a game for free XP and attendance bonuses) and still works perfectly fine using the original Powercolor cooler.
 
I've got a pII 350 genuine Intel on its original Asus p2b board (those genuine ran crazy hot, the celery's were actually cooler) that's @ 18 yrs old now and still works (props to Asus quality). Spent 10 years in Florida and last 10 in my garage. I fire it up every few months. Oh it's also been oc'd to 400MHz it's entire life and spent the first 5 yrs as a server for mIRC chat rooms. It's not heat that kills, its over heating, heat beyond the limits of the weakest component.