570: How hot is too hot?

[Maxwell]

Since I'm impatient, I got myself a 570 to replace the 5970s while they are being RMA'ed. However, I'm a relative CUDA novice. That said, how high can I push the overclock on this card and remain 24/7 stable on a non-gaming rig that is NOT a dedicated crunchher?

Stock clocks for Core/Shader are 780/1560 (shader locked at 2x core clock). I'm currently running at 850/1700 at 65C. Using MSI Afterburner to overclock and monitor temps and whatnot. So how high can I push this without running into issues?

[joker]

Just a guess with my limited experience, but if you have good cooling (probably have to turn the fan up to 100%) I would wager that you could push it to 900MHz once you up the voltage. I got my GTX 470 OC'ed from 607 MHz to 810MHz with fan at 100% running at 66C. As I said, just a guess.

[John P. Myers]

900? pfft. this is Nvidia. if stock is 780, aim for 975, as long as your case can keep it under 85C you'll be fine, although the chips are rated at 105C, there's surely gotta be a degree of life-shortening by approaching that.

[joker]

Fine JPM, I will bite! Ok, so bumped up the clock speed to 853.4 and upped my voltage to 1.0750 and things seem to be running stable for the last hour. Temp at 73C. Update Max??

[DrPop]

Joker, which program are you using to volt mod the GTX 465? I should try doing that to mine. How much fan are you running to hit that 73 temp?

[joker]

Joker, which program are you using to volt mod the GTX 465? I should try doing that to mine. How much fan are you running to hit that 73 temp?

Just MSI Afterburner with fan running at 100%.

[YoDude9999]

I've been running 3x9800GTSCs in my Rodan system for a couple years now.
They get the usual dust bunnies built up in the heatsinks and of course every now and then, I have to clean them out so they can breath again.

Well I've cleaned out two of them within the last few months, but I have one that's been somewhat problematic as it lives in the bottom slot and I have a case fan that blows INWARDS right in that location. Because of this, the heatsink gets clogged very quickly and requires more attendance than the others.

One day, we had a warm day, around 75 degrees outside. This was just a few weeks ago. When I came home from work, I was shocked to see that Everest (Now refered to as, "AID64") was reporting that card as running at 99% activity (along with the other two cards) and its temp was 136c degrees!!!!! AND IT WAS STILL CHUGGING ALONG QUITE NICELY!

Of course when I saw this, I was amazed and took the card offline to keep it from catching fire, but the point being, all the hype about keeping your cards below 60c is probably just hype. This minute, that one GPU card is running at 102c and it's been running at that temp for about three weeks now.

I've run my CPUs at over 100c with IntelBurntest and they still live today...and yes, when it gets really warm out, they crunch at that temp for weeks on end and they still live. I don't get problems with the CPUs until they reach about 112c and then I get BSOD. I just reboot the system and all is well.

Just my two cents worth!

Yo-

[JerWA]

Most of the GPUs are rated for around 100C constant operation. That doesn't mean that's good for longevity (which no manufacturer wants anyways, right? If it works forever you don't buy another one).

Plus at that temp it's pouring massive amounts of heat into your system, cooking things NOT rated for 100C constant operation. I've seen plenty of burst caps and cooked graphics cards. They are definitely not immune to heat.

I'm also surprised you haven't killed your CPU yet. A lot of them will separate the die if you run them that hot for more than a few seconds. Hell if you really run them that hot that long they probably are already dead, and if you took your cooler off the CPU would literally fall apart.

[YoDude9999]

Dude, if you actually think my hardware is dead because of heat issues, I'd suggest you look at my stats, then think again. Don't get me wrong here, this isn't something that I would EVER say to someone, that it's ok to run your CPU and GPUs as hot as I have. I would try to encourage them to invest in better cooling options to keep their hardware at a more sane level. I'm just stating that this is what I've personally seen.

All, and I mean all and every one of my CPUs are running and crunching something. Usually Aqua for them. All my GPUs are also crunching something. The ONLY reason my stats have dropped recently is because of DNETC going down and otherwise all systems are go! AND STILL GOING!

[YoDude9999]

In all honesty, I am truly amazed that I haven't suffered some sort of serious hardware failure by now. With CPU temps of 110c, one would think the transistors would go into "thermal runaway" and just dissolve themselves into silicon goo. Yet, this has not happened and I can tell you that this is true of more than one CPU chip. I have an E8500, which is currently offline because my Q9650 replaced it, but it also has run at temps as this. Additionally, my i7 920 has also run at the 100c level with no apparent damage.

Now, if we speak of longevity, I cannot say for sure how much time they have left in them. Then again, if they only last for 3 years, isn't it time to consider a new box anyway?

Yo-