I'm very confused

[SaltySpitoon]

So anyway, I'm in the process of setting up my first mining rig, after finally finding a place to put one. And I was talking to a friend about what he was running, and he has a few cards, 2 6950s, some 6870s, a few 5770s etc, and I was asking him about his rig before I set mine up, and I was quite alarmed when he told me his have been running at 103C for the last 3 months 24/7. How does that even happen? Don't GPUs die at 80C or so?

Just wondering mostly because this kind of thing majorly effects how I set up.

[1714877363]

1714877363

[1714877363]

1714877363

[1714877363]

1714877363

[ssateneth]

So anyway, I'm in the process of setting up my first mining rig, after finally finding a place to put one. And I was talking to a friend about what he was running, and he has a few cards, 2 6950s, some 6870s, a few 5770s etc, and I was asking him about his rig before I set mine up, and I was quite alarmed when he told me his have been running at 103C for the last 3 months 24/7. How does that even happen? Don't GPUs die at 80C or so?

Just wondering mostly because this kind of thing majorly effects how I set up.

These temperatures are perfectly normal. There are safeguards within a GPU to prevent it from getting to dangerous levels (i.e. it will throttle back to a smaller speed automatically until temperatures are acceptable). Mind, 103C at 24/7 is probably on the high end, but nothing that high end/dual GPU cards won't be subjected to under heavy 3d environments such as cutting edge gaming at super high resolutions

[rjk]

I wouldn't call them perfectly normal, but I recall Graet of Ozcoin trying to kill a card that had a dead fan by mining on it, but I don't remember how long it took to die.

I would aim for 60 to 80 degrees, more than that will accelerate the time it takes for them to die.

[Fiyasko]

So anyway, I'm in the process of setting up my first mining rig, after finally finding a place to put one. And I was talking to a friend about what he was running, and he has a few cards, 2 6950s, some 6870s, a few 5770s etc, and I was asking him about his rig before I set mine up, and I was quite alarmed when he told me his have been running at 103C for the last 3 months 24/7. How does that even happen? Don't GPUs die at 80C or so?

Just wondering mostly because this kind of thing majorly effects how I set up.

Woah!!!!
WOAH WOAH WOAH.
Your buddy IS going to blow his video cards if the gpu's are above 100°C  He must have some swag gear if it hasnt blownout after running like that for 3 months.

90° Is a fine "max" Personally i feel that 95° is when a gpu starts to take damage, But that all depends on the cards themselves.

Just because your buddy's ARE and CAN run that hot, Does NOT mean anyones can, Or that you should even get NEAR there.

Fucking HELL he is Seriously going to blowout his hardware, Odds are that his fans are at 100%, GPU/CPU Fans are NOT meant to run at 100% and you will break them if you constantly run them that fast, I found out the hardway. Trust me. Case fans etc, Goahead, Crank those up.

To reduce temperatures, Drop your memory clock, You'll get HIGHER hash rates (by upto 20mh/s a card, And no im not fucking with you) and colder temperatures.

I Heavily reccommend a $20 20x20'' BOX FAN air circulator, Remove your side panel, And slap that thing on at full speed,
Brought my 6990's 104°C fans speed 100% Way down to 81°C and 71% fan speed.

DO NOT RUN YOUR GPUS THAT HOT, JUST BECAUSE HE CAN, DOESNT MEAN that you or ANYONE else can, or should do that

[Fiyasko]

So anyway, I'm in the process of setting up my first mining rig, after finally finding a place to put one. And I was talking to a friend about what he was running, and he has a few cards, 2 6950s, some 6870s, a few 5770s etc, and I was asking him about his rig before I set mine up, and I was quite alarmed when he told me his have been running at 103C for the last 3 months 24/7. How does that even happen? Don't GPUs die at 80C or so?

Just wondering mostly because this kind of thing majorly effects how I set up.

These temperatures are perfectly normal. There are safeguards within a GPU to prevent it from getting to dangerous levels (i.e. it will throttle back to a smaller speed automatically until temperatures are acceptable). Mind, 103C at 24/7 is probably on the high end, but nothing that high end/dual GPU cards won't be subjected to under heavy 3d environments such as cutting edge gaming at super high resolutions

Yes these safeguards most definitly exsist, Sometimes they work so great you never notice them, Other times they work so great that your GPU just resets on you with a warning (love that feature, honestly)

But what Cannot be measured is the temperature of the fackin capacitors on your GPU, Capcitors a generally rated for continus use at (im not sure, Someone poke me on this part) 60°c?
Altough there maybe some tech unknown to me about monitoring temps on cards that came with pci-e 3, So dont bite at me

[TheHarbinger]

103C?!?!?  Holy crap.

I keep mine at 70C and a variable fan speed not to exceed 80%, after that they start to downclock themselves.

[crazyates]

Gpu temps can be weird. I would consider 80+ to be hot, and 90+ to be dangerous. Now 3 things you need to consider with temps that high:

1) As weird as it sounds, if you're gonna run cards that hot, it's better to do it 24/7, than, say, 20 hours a day. Getting them hot and staying there is better than hot -> cold -> hot -> cold -> etc.

2) Usually VDDC temps are going to be higher than Core Temps. You do NOT want to burn out your voltage regulators. That WILL be what kills your card running that hot 24/7.

3) With temps that high, you still wanna make sure you don't run your fans too fast. That sounds weird, but there is very little to be gained from pushing your fans from 80% to 100%, esp if you have limited airflow already. All you're doing is wearing your fans down faster. 3 months @ 100%? They're due to die anytime now.

[crazyates]

103C?!?!?  Holy crap.

I keep mine at 70C and a variable fan speed not to exceed 80%, after that they start to downclock themselves.

That's almost exactly what I use for settings. 75-80% fan speed, depending on the card.

[crazyates]

But what Cannot be measured is the temperature of the fackin capacitors on your GPU, Capcitors a generally rated for continus use at (im not sure, Someone poke me on this part) 60°c?
Altough there maybe some tech unknown to me about monitoring temps on cards that came with pci-e 3, So dont bite at me

No, but my 5870 has 9 different temp sensors on it. I just checked with GPU-Z. Gives you a pretty good overall picture. Right now, my core may only be 68, but my VDDC is 76.

[SaltySpitoon]

Alright, this is what I had thought, but the last time I had looked at building a mining rig was a few years back, so I thought perhaps the technology had changed.

He said his fan speeds are 45-50%, and one of his GPUs is water cooled. But like a real pro, when he says water cooled, he really means water, not the coolant liquids that you are supposed to use.

[scifimike12]

Note that some cards throttle at lower temperatures.  I remember my 7970's artifacting when they got over 78°C in BF3.

[AzN1337c0d3r]

My 5830 reaches 95C and then starts throttling itself. I have no idea how he got his 5970 to 103C.

[Fiyasko]

My 5830 reaches 95C and then starts throttling itself. I have no idea how he got his 5970 to 103C.

All cards throttle at a different temp, My 6990 will throttle at 105°

[DrG]

My 5830 reaches 95C and then starts throttling itself. I have no idea how he got his 5970 to 103C.

My Double-D XFX 6950s throttle at 103C.

[BCMan]

 So where core temp and vddc temp here, anyone knows? 62 is gpu, 68 - vddc? And what's the 3rd temp?
 

[zvs]

one of my 5970 fans "died" (it was still moving a little bit, it reported 800rpm or something) while I was asleep & when I went in to check on it, GPU #1 was getting about 70mhash/s (and got 30 "invalids", of which I'd had zero in like the month prior), gpu #2 was getting around 150mhash/s (no invalids).  gpu #1 was ~103-104oC, gpu #2 was ~101-102oC. 

i used to adjust the fans up and down (day/night) but I've gotten lazy recently so I just leave them at 80% 24/7.  during the day w/ 90-95o temps and heat index 100o+, they max out at around 72-74oC (running at 810/150 @ 1.010v).  i  guess room ambient temp is probably around 95o also.

anyway....  yeah, i was shocked to see it still 'running' at >100oC.   tbh, I thought they shut down/crashed between 85-90oC.  I don't think I've ever seen a temp >80oC until then..


ah, well.  i took the card apart, blasted the fan with the kitchen sink spray hose, let it dry, and now it's running again, about 200rpm less than the other 5970 in the same system.... and making those noises that one would associate with a fan that's dying, haha.  i'm leaving it going though, cause my replacement isn't due here until thursday =p

[chungenhung]

GPUs can run at close to 105C. They won't die at 103C.

[ssateneth]

GPUs can run at close to 105C. They won't die at 103C.

Pretty much what I was saying in my post

[multi#lord]

So where core temp and vddc temp here, anyone knows? 62 is gpu, 68 - vddc? And what's the 3rd temp?
 

The three temps listed are on specific locations, though they vary depending on the graphics chip/card: they are generally the GPU shader temp (usually highest temp), GPU thermal diode temp, and some other "architectural unit" on the chip/board (including VRMs).

I more accustomed to HWiNFO64, because they usually list what those temps mean.

[BCMan]

So where core temp and vddc temp here, anyone knows? 62 is gpu, 68 - vddc? And what's the 3rd temp?
 

The three temps listed are on specific locations, though they vary depending on the graphics chip/card: they are generally the GPU shader temp (usually highest temp), GPU thermal diode temp, and some other "architectural unit" on the chip/board (including VRMs).

I more accustomed to HWiNFO64, because they usually list what those temps mean.

Yep, sensors now named more informative with HWiNFO. Thanks for the app.