Heat VS voltage

zappian · March 4, 2015

So I am running my 270x at these settings:

A friend of mine started ranting at me saying my voltage was "insane" and that I was gonna brick the card.

I told him the temps were under 68 C and it was fine mostly , I have run a 5850 at 85 C constantly for two years and the card still works fine after 5 years of usage(Damn AMD stock cooler tho).

He said that voltage , even with normal temps , will kill a card super fast.

So what is worse ? High heat on the card say 85 to 90 C , or overvolting it assuming safe temps.

EDIT : @Najuno What do you think ?

Mush Brain · March 4, 2015

I modded my GTX 480 bios and ended up frying a voltage regulator after 4 years...

JAKEBAB · March 4, 2015

Id say voltage would kill quicker over high temps but i don't know the max voltage for your particular card.

zappian · March 4, 2015

I modded my GTX 480 bios and ended up frying a voltage regulator after 4 years...

4 years is fine , a 270x(7870) is a three year old card.

Was released in 2012 , I will want a new card in two years and I will start saving for it.

zappian · March 4, 2015

+44mV is +0.044V

That's only a small fraction of the volt increase when the GPU goes from idle to load.

Tell your friend to go back to school.

I will

I thought the same it was insignificant , I will tell him.

Najuno · March 4, 2015

So I am running my 270x at these settings:

A friend of mine started ranting at me saying my voltage was "insane" and that I was gonna brick the card.

I told him the temps were under 68 C and it was fine mostly , I have run a 5850 at 85 C constantly for two years and the card still works fine after 5 years of usage(Damn AMD stock cooler tho).

He said that voltage , even with normal temps , will kill a card super fast.

So what is worse ? High heat on the card say 85 to 90 C , or overvolting it assuming safe temps.

EDIT : @Najuno What do you think ?

When it comes to OCing there is always Watts and Volts and the ASIC of the chip including the quality of your VRM and memory controller. The ASIC is what people call silicon lottery. For more modern GPUs the ASIC isn't as important anymore as it was in the past, but still determines the purity of the chip and therefore in what class this chip OCs best. That could be: air, aio, custom water, phase change, DICE or LN2.

After determining what kind of ASIC the chip on the graphics card has it is a matter of the right cooling solution of all crucial parts on the PCB and how much phases you have, to make an OC possible.

The GPUs nowadays with aftermarket air coolers (nVidia reference from GTX Titan on as well) are save to OC out of the box as far as you can push them, there's little to no chance at all to fry the card with an OC. They all have save locks in the BIOS that will allow you to raise TDP, PD and Voltage in a moderate way. With tweaking software like MSI Afterburner of EVGA Precision X you can try to hit the BIOS available maximum but worst case is a BSOD or driver crash in Windows and the card will clock itself back to stock. Even when your cooler runs (does not apply for the old reference coolers!) on maximum and your OCing is high because you have a good sample, worst case in benchmarks or games will be some thermal throttling and the card will clock itself down until the temps hit acceptable areas. This is i.e. why the R9 290 and R9 290X were so limited with the reference coolers in their performance. They ran into very bad throttling issues. Of course there were other issues as well, that got fixed by the board partners later very good (MSI Lightning R9 290X beautiful example) but I don't want to go in depth with that now.

As for your OC, when the card is benchmark stable in short term stress test and long term loop test, and you do not run into thermal throttling, it is save to say you achieved a good and stable OC. The best test programs are 3D Mark FireStrike for short term and Unigine Valley 1.0 for loop or Unigine Heaven 4.0 for loop, but with Heaven and higher OCs it is likely that air coolers run into thermal throttling, because Heaven is very demanding and more for custom watercooled cards to be run in loop to test the stability of the extremer OC and Temperature Delta.

Does OCing reduce life time of your hardware? Yes it does, but very much depending how crazy and hard you OC. The manufacturer of GPUs just have to guarantee the clock, boost and VRAM speeds that are written on the box. That means if the card is not able to hold that stable in endless loop, that is a RMA reason. With OCs with a standard BIOS you will not significantly lower the life span of a GPU because the TDP, PD and Voltage stay in a very moderate limit. As example here I take one of my GTX980s. It is a reference card from nVidia that OCs with air and the standard BIOS up to 1552MHz boost stable in benchmarks. With all sliders maxed out TDP, Temp and Voltage. Pretty good hm? No! I write my own BIOS and I am able to achieve under air more than 1700MHz with a no limits BIOS BUT this BIOS can actually destroy the card easily, so I have to use it with care. I would never use that on a daily base for gaming, this would shorten the life span of my card dramatically. But as long as I stay within the standard BIOS I will shorten the life span just a little bit, and that little bit is so little, that the card is outdated at the time it dies, that it isn't even used anymore in HTPC like 2018.

Is it even worth to OC? Yes of course it is! Considering you have a good sample that is able to OC stable why not? It is performance for free. It isn't only for benchmarks, statistics and e-peen. I have tested it with my GTX980 in single run in 1440p. It makes almost a performance difference in average in games of 17%. 17% more performance for free is something that I consider to be worth.

Most GPUs start, of course based on their BIOS, thermal throttling at 95 degree Celsius in order to prevent the chip and VRM and VRAM from burning out. And they can take beatings with higher temps for hours. Sure heat is the main enemy of every electronical part so the lower you are in temp the better it is. But I run my GTX980s with air cooling with slot1 PCIe at 86 Celsius with OCs without any issues. 85 Celsius in your case are fine and I personally wouldn't be worried about it too much. But keep in mind every piece of hardware is different, even the same brand, make and model is different. If you feel comfortable with it keep it.

I hope this helps you understand all that temp and Voltage relation a bit better. Of course this is a never ending topic because I didn't really go in detail with many features or cooling, but this is already a wall of text .