The only way I would be comofortable at temps in the 80s, 90s, or even 100C like that is if they were going to be run at those temps for 24/7 for long periods of time.
I run dedicated linux miners 24x7. I have a separate machine for gaming.
Topic: Safe temperatures for GPU (Read 43815 times)
I run dedicated linux miners 24x7. I have a separate machine for gaming.
I keep my temps on my 7970 pretty low, as I'm often turning it off and on again as I game or go back to mining. Varying the temps from 40C to 70C is acceptable. The only way I would be comofortable at temps in the 80s, 90s, or even 100C like that is if they were going to be run at those temps for 24/7 for long periods of time. Running them from 100C to 30C and back up to 10C every 12 hours will kill your cards faster than just running them at 100C.
yep, I've not overclocked them. this is stock. as soon as I get a stable OS where I can run multiple cards I'll start to tune them. Nice to see you getting 300, that's what I expect.
No adverse affect? Holy shit what cards are you running?
several reference design HD6870x2's with dual gpu's and heat pipes.
Code:
GPU 0: 89.0C 425531RPM | 251.8M/251.4Mh/s | A:1341 R:0 HW:0 U: 3.56/m I: 4
GPU 1: 101.0C 100% | 280.1M/279.9Mh/s | A:1507 R:2 HW:0 U: 4.01/m I: 4
GPU 0: 91.0C 992647RPM | 252.2M/252.2Mh/s | A:1323 R:1 HW:0 U: 3.52/m I: 4
GPU 1: 99.0C 100% | 279.7M/279.7Mh/s | A:1544 R:0 HW:0 U: 4.12/m I: 4
just chugging along.... they actually seem to run better when they are hotter
I read that most AMD spec cards can handle 120 celcius but I haven't seen a good source for information from actual manufacturers, most of it is just heresay and it's based on CPU temps. Don't get me wrong, If I ran these in windows they might show artifacts. All the better reason to send them back after 700 days. I'm performing a testing service for the manufacturer.
I wonder if you've just prooven to me that a Hot not overclocked card runs steadier/safer/better than an overclocked cold card....
i try to stay below 80c
No adverse affect? Holy shit what cards are you running?
several reference design HD6870x2's with dual gpu's and heat pipes.
Code:
GPU 0: 89.0C 425531RPM | 251.8M/251.4Mh/s | A:1341 R:0 HW:0 U: 3.56/m I: 4
GPU 1: 101.0C 100% | 280.1M/279.9Mh/s | A:1507 R:2 HW:0 U: 4.01/m I: 4
GPU 0: 91.0C 992647RPM | 252.2M/252.2Mh/s | A:1323 R:1 HW:0 U: 3.52/m I: 4
GPU 1: 99.0C 100% | 279.7M/279.7Mh/s | A:1544 R:0 HW:0 U: 4.12/m I: 4
just chugging along.... they actually seem to run better when they are hotter
I read that most AMD spec cards can handle 120 celcius but I haven't seen a good source for information from actual manufacturers, most of it is just heresay and it's based on CPU temps. Don't get me wrong, If I ran these in windows they might show artifacts. All the better reason to send them back after 700 days. I'm performing a testing service for the manufacturer.
90+: about to get throttled
80-90: hot
70-80: safe
<70: cool
80-90: hot
70-80: safe
<70: cool
lol.
I've been running several ATI cards at 100 degrees Celsius 24x7x365 for YEARS with no noticeable effect, they just keep right on mining along. There is little question that driving a card this hard will shorten it's life but unless I spray water on the chip and actually crack it the functionality I need remains fully intact. I've had them up to 105c in the summer. I can't find it but the rumor is you can safely operate an AMD GPU up to 125 degrees. Gold melts at 1000 degrees C, I think we'll be fine.
Run em hard... the warranty is usually only two years so if you can break them before the warranty is up you get a nice new replacement. ;-)
Its Extremely card specific about those operating temperatures, I dont know of ANY !ATi! card that can operate at >100c°
The hottest Functioning card i had was a crossfire dual AMD MSI 6990's that ran at 104°C and would overheat and crash at 106°C And i had a giant box fan trying to cool them!
No adverse affect? Holy shit what cards are you running? I had to drop the core on my 6990's by 10mhz after a month otherwise they wouldnt function, After another three months one of them had a GPU core blow out, It Died a week later, Then the second 6990 also had a GPU core blowout, a month after the first one blew out..
I am running 7950s as well. >80 is too hot, and you're going to see less efficiency in regards to hashing effectiveness.
Get SpeedFan and check what your GPU fans are set to for speed. Mine were set default @ 50%, which was getting temps in the mid-80s. I changed this to have them run at 80%, and now they're averaging 65-75*C, which is much, much better.
Get SpeedFan and check what your GPU fans are set to for speed. Mine were set default @ 50%, which was getting temps in the mid-80s. I changed this to have them run at 80%, and now they're averaging 65-75*C, which is much, much better.
90+: about to get throttled
80-90: hot
70-80: safe
<70: cool
80-90: hot
70-80: safe
<70: cool
lol.
I've been running several ATI cards at 100 degrees Celsius 24x7x365 for YEARS with no noticeable effect, they just keep right on mining along. There is little question that driving a card this hard will shorten it's life but unless I spray water on the chip and actually crack it the functionality I need remains fully intact. I've had them up to 105c in the summer. I can't find it but the rumor is you can safely operate an AMD GPU up to 125 degrees. Gold melts at 1000 degrees C, I think we'll be fine.
Run em hard... the warranty is usually only two years so if you can break them before the warranty is up you get a nice new replacement. ;-)
90+: about to get throttled
80-90: hot
70-80: safe
<70: cool
80-90: hot
70-80: safe
<70: cool
I have a rig setup with two MSI AMD 7950s. I am running at the default clock speeds... not overclocking anything. The temerature of the GPUs are around 84C. I have the case open to let the heat escape. Is it safe to keep this setup running 24x7 ? What is a safe point and when do I need to worry?
What can I do to improve the temperatures? I am planning on putting a large fan that will blow air into the GPU cards.
What can I do to improve the temperatures? I am planning on putting a large fan that will blow air into the GPU cards.
If your mining bitcoins, Drop your memory clock alll the way down, You'll save about 3-6°C
I run my 7950 at 75-78°C 24/7 And even at that im not too happy (fan at 80%) So long as you keep the temps under 82°c Im sure mining 24/7 at that temperature will be fine
one of the more important things is to keep your fan AWAY from 90%+ speed, This will reduce your gpu life Extremely
24/7@85% is about the highest reccommended
I'd say this is OK for the GPU (although I feel more comfortable with <80C for 24/7 use) but what are the temperatures of your VRAM and VRMs? Use GPU-Z to check all of the card's sensors.
It seems that GPU-Z is only for Windows. Is there anything comparable for Linux (Xubuntu) ?
i would put a fan in front and at the rear off the cards and check if this helps.
offtopic : did you try bitcoin mining on your rig if yes how mutch did whas your hash rate
offtopic : did you try bitcoin mining on your rig if yes how mutch did whas your hash rate
6 - 12 month? most cards will still be covered by warranty 
I am only comfortable when temp < 75, using extend cable does help as well as running the miner naked.
I am seeing the temp difference in/outside the case is abou t 5-6 degree, and the extend cable connected cards are running also 5-6 degree less, so that adds up about 10 degree difference.
I am only comfortable when temp < 75, using extend cable does help as well as running the miner naked.
I am seeing the temp difference in/outside the case is abou t 5-6 degree, and the extend cable connected cards are running also 5-6 degree less, so that adds up about 10 degree difference.
Take into consideration the thermal throttling speed, which may be around 93C for 7950s. Keep as far below as is reasonably possible, and just know the cooler you can keep them, the longer your investment will last.
There is nothing WRONG with going up to 90C or staying there, but it is decreasing life span.
There is nothing WRONG with going up to 90C or staying there, but it is decreasing life span.
I am using this for mining Litecoins. The useful life of this hardware will be 6-12 month max, I guess. So, if it can last that much, it's ok. I am going to install a couple of fans to blow air at the GPUs and see how low the temp will go.
They can run that hot, but in general, cooler is better! I try to keep mine at ~70C, 75C tops. When I was undervolting, I really liked having them temps down into the 55C range. Don't be afraid to crank the GPU fan (I run mine at 80%), and open up your case or do whatever you can for some increased airflow.
Take into consideration the thermal throttling speed, which may be around 93C for 7950s. Keep as far below as is reasonably possible, and just know the cooler you can keep them, the longer your investment will last.
There is nothing WRONG with going up to 90C or staying there, but it is decreasing life span.
There is nothing WRONG with going up to 90C or staying there, but it is decreasing life span.
I'd say this is OK for the GPU (although I feel more comfortable with <80C for 24/7 use) but what are the temperatures of your VRAM and VRMs? Use GPU-Z to check all of the card's sensors.
anything under 90c is normal/fine for GPUs
I have a rig setup with two MSI AMD 7950s. I am running at the default clock speeds... not overclocking anything. The temerature of the GPUs are around 84C. I have the case open to let the heat escape. Is it safe to keep this setup running 24x7 ? What is a safe point and when do I need to worry?
What can I do to improve the temperatures? I am planning on putting a large fan that will blow air into the GPU cards.
What can I do to improve the temperatures? I am planning on putting a large fan that will blow air into the GPU cards.
Jump to: