Bitcoin Forum

What is a good temp?

My rigs are running 24/7 in the basement, each card hovering around 60C

I like to keep my miners running all the time so I can reach over and light a cigar off of the nearest video card.

Here's a pro-tip: A little potpourri in a cup on your video card is really just a nice thing to smell and it helps cover the scent of burning electronics! And when a power supply blows a capacitor, bam!, free fireworks for the kids. Now you don't have to take those freeloaders to a movie. Also, kids love to take turns sitting and watching daddy's computers, clutching their little hands around a fire extinguisher while daddy sleeps.

As a bonus, you'll know exactly how much power you can sustain through an ordinary extension cable as it runs from a handmade light socket power adapter plugged into your neighbor's conveniently-located and lighted garden shed. As another bonus, it turns out that annoying neighbor cats don't, in fact, have nine lives.

As for having computers running nonstop in your bedroom, really this is just a special treat. Use that hot, hot video card to keep the KY tube from hardening. If your case has enough room, this is also a good place to keep the old fleshlight warm and ready to go.

Really, keeping a miner running around the clock is a good alternative heat source for me, as my normal process involves taking "donations" of furniture from the nearby goodwill. And the loudness of the fans help drown out the screaming. Also the gunshots. Also the screaming.

^ I concur.

My single 5970 runs at 850/300 fans 92% with a small fan blowing air directly between the gpus. 68-72c all day.   

temp near 80 24/7 2 months
4 cards

Cards @ a higher voltage/clock seem particularly susceptible to becoming unstable when temps rise. @ 1.18v, 890MHz core, my 5850s have no problems running up to 100c if I try to conserve energy by turning the fans in the room down or off during hot summer days. @ 1.198v, 980 MHz core, my 5850s are prone to the graphics drivers failing once temp rises above 80c.

At any rate, while heat may lessen the lifespan of a GPU if constantly above 90c (which probably won't be a concern to you if you're using software tools to bump voltage/clocks up and your card has a 3-year+ or lifetime warranty as many do), it's perfectly safe to run cards 24/7 assuming the electrical lines running to your rigs is rated a safe amount above what you're demanding of them (don't neglect this!). Do ensure your fire detectors are functional and.... in existence... Can't ever be too careful (well... until it becomes too costly, anyway  ;) ). Investing in some fire extinguishers and fire insurance may be wise, too, but if the environment is dry, this shouldn't be an issue (though there have been some cases of power supplies exploding, but this isn't exclusively because they were run 24/7).

@horkabork: LMAO

I try to stay below 85°

A script to shut down the miners and eventually the rig if temperatures go above 89°C gives me a better feeling about 24/7

Wouldn't the GPUs automatically throttle down if it gets too hot?

yes, but way late ~105°

also there are some other chips that often are hotter than the gpu. in gpu-z you can see some more measuring points if you are lucky with your card.

LOL, love it  ;D

GPUs throttle or shut off when they're at such a high temperature that the components are in immediate physical danger. That point is usually way higher than the temp at which the card becomes unstable.

The card I´m using is not overclocked at all or anything like that. Though I guess I shouldn´t have any problems letting the miner run 24/7?

Btw...what about a non overclocked CPU? Any problems if it stays on 99% 24/7?

What is your criteria for problems? Running a card at 100% usage 24/7 will almost certainly degrade the card. Now if it degrades its useful lifespan from 6 years to 4 years, that's not a real problem for most people, but there's no way to know any of that in advance.

Basically you are murdering your card, but you are doing it slowly. Figure out for yourself what that means to you.

two words: box fan

Mine have been running at 90-95 degree celcius for over 10 days straight. I really don't think temps damage a card unless their heading over 100 degree celcius. Cards tend to scale back processing as they hit that temp so there's a built in failsafe. Or they might blow up.

Share the script? :)

oh man i just died! +100 BTC credit for you!

10 days of runtime on a few cards is an incredibly poor sampling to draw from for statistical analysis. Heat is a proven killer of hardware, no ifs ands or buts about it, the rate varies however.

In all likelihood the cards will still last for a few years even running quite hot 24/7, but you certainly increase the chance of hardware degradation and failure, and the rate at which it heads to that. The MTBF is likely the length of the original warranty on the card, estimate perhaps 10-30% shaved off that.

It's safe up to a point. My 6990's are below 80 most of the time and have been for months. 5850's stay 75ish. But I think more important than just the heat is the electricity usage and fire hazard.

I have a rig that pulls 11 amps when mining. I purchased one of these Kill-a-Watt's (http://amzn.to/jBt0rJ) and tested each of my setups first. Then I went to my breaker box and figured out how many circuits there were and if I needed to move anything to another outlet.

Turns out my 11 amp rig is on a 15 amp circuit. So I moved everything else off it and I am just below the 80% recommended limit for 24/7 usage. Then I purchased renters insurance ;)

My temps on my 6950 are 40° all the time! OCed to 944MHz on stock volts. I am water cooled though! ;D My 5850, and 5770 on the other hand, aren't so lucky. They are both @1GHz, and both run between 65-70°, on air, in other computers. Ambient temps are about 18°C. They're all running till they die. I'll let you know how long that is! :)

24/7 without right cooling is suicide for rigs. Unless you do it right, you will start to destroy your 6990 5970s. Single gpus aren't so much the prob. I highly suggest you do proper cooling with dual GPU cards or you can say good bye to one of the two gpus soon.

maybe I will add something to clocktweak in a week or two when I have the time...

Hahaha, made my day ;)

I accidentally knocked out a couple of my miners from the outlet today.

Was only 800w power draw, but severely melted Kill-A-Watt and an outlet-splitter.

Both were rated to handle over 1.5kw.

Just a heads-up. Stay safe, friends.

I was looking at my power box yesterday at my shop which I am setting up at.  I have a spare 70A 240V breaker (also a spare 100A 3-Phase).  My shop has Industrial 3 phase I am guessing to be at least a 300A main line...given area averages I'm looking at less than .05/kwh.

So I think I'm good on power.  Built a box which has a boxfan with a filter blowing in and a boxfan without a filter blowing out.  Starting with 3 rigs ~3.1gh/s. 

Now if I can only find more 5830's.


But on the question of 24/7...why wouldn't it be safe...just don't get greedy...no need to exceed the temp rating for that extra 1btc a month.  Do the math on the cards/processor/mobo current use...add 10%(or more) and that is the psu you need.  I think the main risk is drawing too much current, most modern PSU's can run on 240v also, if you wire into a 240v line you cut your current in half...be sure not to overload your breakers.