How hot is too hot - GPU


log in

Advanced search

Message boards : Number crunching : How hot is too hot - GPU

1 · 2 · Next
Author Message
Profile BigWaveSurfer
Send message
Joined: 29 Nov 01
Posts: 166
Credit: 8,415,261
RAC: 9,208
United States
Message 1278187 - Posted: 31 Aug 2012, 13:52:09 UTC

I'm curious around what temp people get concerned about their GPU? Currently my AMD shows it is at 77C (172F), to me that is hot, but the fan speed is set to auto and only on 33%. I hit manual and pushed it to 100% and she quickly dropeed into the 50's. At what point to I need to be concerened? Activity on the card is jumping from around 65% to 80%.
____________

Profile MikeProject donor
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 24510
Credit: 33,845,675
RAC: 24,195
Germany
Message 1278197 - Posted: 31 Aug 2012, 13:59:20 UTC

Everything below 80°C is fine.
Mine is running with auto settings at 51% fan speed and 71°C

____________

Profile ignorance is no excuse
Avatar
Send message
Joined: 4 Oct 00
Posts: 9529
Credit: 44,433,321
RAC: 0
Korea, North
Message 1278215 - Posted: 31 Aug 2012, 14:29:58 UTC - in response to Message 1278197.

I keep my fan speed at 65% and it keeps the GPU in the high 40's to mid 50's.
____________
In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

End terrorism by building a school

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5864
Credit: 60,523,140
RAC: 47,623
Australia
Message 1278499 - Posted: 31 Aug 2012, 23:14:10 UTC - in response to Message 1278215.


My GPU generally sits around 70°c, the fan speed ramping up automatically as the ambient temperature increases.
____________
Grant
Darwin NT.

AlProject donor
Send message
Joined: 3 Apr 99
Posts: 481
Credit: 53,480,057
RAC: 29,394
United States
Message 1278509 - Posted: 31 Aug 2012, 23:47:31 UTC

A reply not based upon any scientific data that I can reference, but I have to think that long term, cooler=better, regarding the expected longevitiy of the product.
____________

Josef W. SegurProject donor
Volunteer developer
Volunteer tester
Send message
Joined: 30 Oct 99
Posts: 4302
Credit: 1,070,845
RAC: 1,128
United States
Message 1278514 - Posted: 1 Sep 2012, 0:16:41 UTC - in response to Message 1278509.

A reply not based upon any scientific data that I can reference, but I have to think that long term, cooler=better, regarding the expected longevitiy of the product.

You're right, the rule of thumb is that a 10° C reduction about doubles the life of electronic parts (based on the Arrhenius equation). OTOH, the card dies if the fan fails so max cooling might actually decrease longevity.

Keeping the card alive until any user would consider it obsolete can probably be done with compromise settings. But I do wonder if the automatic fan settings may be somewhat on the slow side both because many users are interested in quiet and because the manufacturer's idea of when the card will be obsolete may be sooner than many user's idea.
Joe

.clair.
Volunteer moderator
Send message
Joined: 4 Nov 04
Posts: 1300
Credit: 23,063,564
RAC: 596
United Kingdom
Message 1278565 - Posted: 1 Sep 2012, 2:30:54 UTC

I like to keep mine under 70c
because its a number i picked out of the air a while ago coz it seemed about right,
no other reason for it.

Profile ignorance is no excuse
Avatar
Send message
Joined: 4 Oct 00
Posts: 9529
Credit: 44,433,321
RAC: 0
Korea, North
Message 1278641 - Posted: 1 Sep 2012, 5:05:30 UTC - in response to Message 1278514.

A reply not based upon any scientific data that I can reference, but I have to think that long term, cooler=better, regarding the expected longevitiy of the product.

You're right, the rule of thumb is that a 10° C reduction about doubles the life of electronic parts (based on the Arrhenius equation). OTOH, the card dies if the fan fails so max cooling might actually decrease longevity.

Keeping the card alive until any user would consider it obsolete can probably be done with compromise settings. But I do wonder if the automatic fan settings may be somewhat on the slow side both because many users are interested in quiet and because the manufacturer's idea of when the card will be obsolete may be sooner than many user's idea.
Joe

a replacement fan for an ATI/AMD GPU is $15 online. An aftermarket HSF much quieter than the original costs around $50 A gpu is much more expensive. I'd rather burnout a fan than the GPU
____________
In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

End terrorism by building a school

jravin
Send message
Joined: 25 Mar 02
Posts: 941
Credit: 102,407,679
RAC: 88,049
United States
Message 1278831 - Posted: 1 Sep 2012, 12:29:37 UTC - in response to Message 1278641.


I'd rather burnout a fan than the GPU


True! but how do you catch a dead fan before the GPU burns out, too? I would think that it would only take a few minutes at most before the GPU itself goes up in smoke....

____________

Profile BigWaveSurfer
Send message
Joined: 29 Nov 01
Posts: 166
Credit: 8,415,261
RAC: 9,208
United States
Message 1278834 - Posted: 1 Sep 2012, 12:33:18 UTC - in response to Message 1278641.

Joe[/pre][/quote]
a replacement fan for an ATI/AMD GPU is $15 online. An aftermarket HSF much quieter than the original costs around $50 A gpu is much more expensive. I'd rather burnout a fan than the GPU[/quote]

What if the fan burns out when no one is home? will the GPU shut down and prevent it from damaging itself if it gets too hot (without any additional 3rd party software installed)?

Curious if people are noticing their GPU's lasting a certain amount of time before failing (1 year, 2 years)?

On my Nvidia set up I'm not too concerened because I have a lifetime warranty from EVGA on the card, but the AMD card is OEM, although if it does crash I would upgrade it and try to find another one of those warranty deals (not sure if they are still offered or offered on those cards).
____________

Profile Link
Avatar
Send message
Joined: 18 Sep 03
Posts: 828
Credit: 1,571,473
RAC: 262
Germany
Message 1278847 - Posted: 1 Sep 2012, 12:49:33 UTC - in response to Message 1278831.

but how do you catch a dead fan before the GPU burns out, too? I would think that it would only take a few minutes at most before the GPU itself goes up in smoke....

Fans usually don't die "suddenly" out of nothing, they first start to vibrate because of bad bearing, the stronger the vibration, the slower they will run. That means usually you should have few days to replace them, before it gets critical, if you need more time, you can even put little oil on them (in the bearing) and get some days or even weeks more, where the fan runs like new again. Sure, it can happen, that a fan dies suddenly, but I'd say that's less common (at least from my experince).
____________
.

Profile Fred J. Verster
Volunteer tester
Avatar
Send message
Joined: 21 Apr 04
Posts: 3250
Credit: 31,892,990
RAC: 4,189
Netherlands
Message 1278870 - Posted: 1 Sep 2012, 14:24:27 UTC - in response to Message 1278847.

but how do you catch a dead fan before the GPU burns out, too? I would think that it would only take a few minutes at most before the GPU itself goes up in smoke....

Fans usually don't die "suddenly" out of nothing, they first start to vibrate because of bad bearing, the stronger the vibration, the slower they will run. That means usually you should have few days to replace them, before it gets critical, if you need more time, you can even put little oil on them (in the bearing) and get some days or even weeks more, where the fan runs like new again. Sure, it can happen, that a fan dies suddenly, but I'd say that's less common (at least from my experince).


I've cleaned and reseated a GTX470 full of dust and replaced the stock fan of
the Q6600 for a better one, it has ran for 18 month at ~100C!

Well, I believe the actual temp reading is a bit too high, but a measurement with a thermo-coupled device, gives a 95C reading.

So far I only burned a 9800GTX+ and a HD4850 GPU, both when the fan failed,
the FERMI/KEPPLERs and AMD/ATI 5000 series and up, are better protected.
Also use SmartDoctor* to set GPU core-clock a bit lower, 1200MHz in
stead of 1440MHz. *Or EVGA Precision.

The i7-2600+ 2 HD5870 GPUs now runs with one side open and an external
15 inch fan. Since I increased the base-clock from 100MHz to 102MHz, that's
actually 2x 34-38(multiplier)= 68 to 76MHz. DDR3 1632MHz DRAM also gets hot,
85C, 20C more as the 800MHz(FSB=1602MHz) DDR2 DRAM on my 2 NVidia hosts.

The 5870 GPUs were on 88C and the (i7-2600)CPU (stock-cooler) at ~82C with 6
out of 8 cores used(Docking@home), 2 cores are driving the 2 GPUs.

Even doing 1 Einstein and 1 SETI MB WU at a time, each using half a GPU.
Whitout making errors ofcoarse.
I'm more worried about errors at (too) high temps. The up-to-date CPUs and
GPUs appear to be better protected against too high temps.

To protect the mobo from burning, don't draw power from the PCI(e) slots
and/or USB 1.1/2.0/3.0 connectors.(Charching devices from USB ports is not a
good idea).
They all are Multilayer and long and thin connections can really get
too hot and burn, beyond repair.
(The INTEL DP67BG mobo has options to OC, but also can limit the CPU current to
97Amps f.i., protecting the CPU and mobo).

As a rule of thumb, I use 90C* as the max safe temp, preferrebly lower.
*If no errors occur.





____________

Profile Alex Storey
Volunteer tester
Avatar
Send message
Joined: 14 Jun 04
Posts: 551
Credit: 1,666,510
RAC: 561
Greece
Message 1278889 - Posted: 1 Sep 2012, 15:30:40 UTC

For those of you that like to read, but hardly ever (or never) post:

a) You should post!:)
b) Keep in mind that all the above advice is for tower/desktop PCs. Laptops can run a lot hotter and still live a healthy & long life. A crude rule of thumb is if you here your laptop fan going into turbo/lift-off mode, it's time to clean the vents.

Profile MikeProject donor
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 24510
Credit: 33,845,675
RAC: 24,195
Germany
Message 1278919 - Posted: 1 Sep 2012, 16:44:56 UTC

Oh its possible Mark.

I fried one maybe 10 years ago.
You just need to try very hard LOL

____________

Profile Paul D Harris
Volunteer tester
Send message
Joined: 1 Dec 99
Posts: 1123
Credit: 33,598,472
RAC: 0
United States
Message 1278935 - Posted: 1 Sep 2012, 17:20:25 UTC - in response to Message 1278921.

Oh its possible Mark.

I fried one maybe 10 years ago.
You just need to try very hard LOL

LOL.....dammit....I shall have to try harder.

Realistically speaking though, I don't even push the CPUs that hard anymore.
The GPUs do all the real work now, so OCing the crap out of the CPUs is not so much of an issue anymore.

I got my reputation here on Seti back in the day by being able to OC every single clock cycle out of every CPU I owned.

Now, not so much. There are users in my own team that can whack the kitties with a single 690.

Things have changed.


A 690 cost about a grand.
____________

chromespringerProject donor
Avatar
Send message
Joined: 3 Dec 05
Posts: 270
Credit: 22,829,709
RAC: 45,525
United States
Message 1278976 - Posted: 1 Sep 2012, 18:46:07 UTC

just completed dismantle and cleaning of the pc for the up coming winter months .. currently running an XFX R7970 at 1000MHz clock and 1425MHz memory clock @ 100% load with manual fan setting @ 65% and a stabilized gpu temp @ 61C with ambient temp @ 76 F

____________

spitfire_mk_2
Avatar
Send message
Joined: 14 Apr 00
Posts: 458
Credit: 12,818,611
RAC: 8,960
United States
Message 1279021 - Posted: 1 Sep 2012, 21:09:01 UTC

I have nVidia GTX 460 with 2 GB of RAM.

I run 7 gpu work units, it consumes 1.8 GB.

GPU-z shows that:
GPU Temp: 71°C
Fan Speed (RPM): 1500 RPM
Fan Speed (%): 34%

As you can see, if I want to tinker with fan speed settings, I can push the fan speed higher (50 or 60% for examples) and bring the temperature down a few degrees. As is, even though the program reports 71°C, the fan runs at only 34% of available speed.

____________

1 · 2 · Next

Message boards : Number crunching : How hot is too hot - GPU

Copyright © 2014 University of California