GPU Serious Heating Issues

[insider]

Okay my GPU seems to reach 100% usage the instant my game starts and the temperature slowly climbs from 40 deg C to 75 deg C within a matter of 3-4 minutes which is making game testing almost impossible as I recently lost two 2 TB hard drives which just dropped dead one after the other in quick succession and I may have to suspect heat as a contributing factor, I forgot to mention the GPU its an ASUS Nvidia GTX 760, so any advise to keep the temperature down would be great, I used to do that before by manually controlling the GPU fans and setting them to 100% which would make them run at approx 5000 rpm not sure how recommended this is, so any advise would be great.

I had my eyes set on the 980 for a long time, but it was in the market too long so I knew there was a better card on the way and the GTX 1080 sounds impressive although the price is steep, it should serve my purposes for another 2-3 years.

The 3 year warranty on my GTX 760 expires in another 6 months, so I plan to upgrade only after that, so any tips to run my GPU at earthly temperatures. My cabinet has Front, Side, Rear, Top and Bottom fans and all are set to full speed via my fan controller so the only option that I can seem to exploit is making the GPU fans spin to full speed, any recommended tools for that.

And it's awful quiet in here, every time a new series of GPU releases they would be a thread for it, Kojack are you not impressed with the 1080 ?
I am sure cybereality must be angry ( 3 x 980) if you know what I mean

[EricB]

Enable vsync to limit the framerates thus limiting the GPU and reducing temps.


RX series is much more impressive for the price than the 10xx series.

[insider]

Enable vsync to limit the framerates thus limiting the GPU and reducing temps.

Technically it is enabled, but still my GPU goes to 100% that is what is confusing me, is my card faulty, I also noticed a drop of almost 20 FPS with the latest Nvidia driver, so not sure where the issue is.

[Zonder]

I think you should check out your PSU with 2 hard drive breaking as that is a sign that voltages are wrong.

I presume you have removed all the dust from your gpu recently?

[insider]

I think you should check out your PSU with 2 hard drive breaking as that is a sign that voltages are wrong.

I presume you have removed all the dust from your gpu recently?

One of the hard drive's fault seemed very serious it just crashed and wouldn't allow the computer to start up until it was plugged out and the other drive although passed the SMART test, just died after I copied a few important files, so they both crashed with a month's gap, The PSU is a Corsair RM750 and no I haven't opened the cabinet since Feb this year when I upgraded my RAM and no I didn't dust anything at all.

My GPU's idle temperature used to be at 33 when I bought it in 2014 now it's 35 so it's dusty but I am sure not that dusty.

[ppd]

In the last few years HDD reliability has fallen off the cliff. Unofficial quarterly data from Backblaze (last: https://www.backblaze.com/blog/hard-dri ... s-q1-2016/) are extremely useful for avoiding worst offenders. For temperature, older disks could endure 50-55°C for prolonged periods, but for current ones temperature should not exceed around 45°C, preferably under 40°C. Temperatures are available via SMART. Modern desktop cases, with PSU at the bottom and HDD cage low directly behind front fans provide sufficient cooling, but in old style towers HDDs are often mounted high in a heat trap.

[frostbyte]

for sure it's a good advice to check the PSU/board before you plugin your'e unordered future gpu( 2*200$ 5.5tflops RX 480 (-: )

there is a slim chance that my "faulty" nvidia gpu( laptop ) story may be related...
a few month ago i decided to extend one of the partitions with an Unallocated hard drive space,
later at some point ,weird hard crushes/freezes started to appear every time i used the nvidia,
after trying everything i could think of, i just gave up on the nvidia...

long story short...turned out that the unallocated drive had a bad sector, windows decided to use this BAD SECTOR as part of the PAGEFILE
every time the pageFile got big the BAD sector would cause my computer to crush/freeze
for some strange reason only the nvidia GPU produced a massive pageFile( integrated HD was fine )

lesson learned- the problem may not be with the gpu...try testing the gpu with another machine...

[Zonder]

I think you should give it a clean in general hold fans when you clean them then check temps as it's simple to do before you do anything dramatic

Make sure no cables are near your GPU as well tbh there should only be the power cable near and that should be away from it anyway.

Funnily enough my corsair psu (HX1050) just died recently loud bang and and flipped the breaker to my house! and have to say it did protect the machine from damage so you get what you pay for and they are replacing it under warranty it's 3 years old they don't make that any more so they are replacing it with the HX1000i so can't complain

[insider]

There is a good chance my card is faulty, whenever I open Chrome nowadays there is a small black line at the bottom and the webpage doesn't render, I have to close Chrome and then reopen it and then it works fine, today I was checking out a Youtube page and suddenly the webpage started flickering and yes I have hardware acceleration enabled and Chrome is at is default settings and fully updated but the point is the GPU is exhibiting strange behaviour one by one so there is a good chance it is faulty but these are random issues I can't claim warranty unless I can prove the card is malfunctioning.

@Zonder Yes I have dusted it on Friday, idle temperature is down to 33 from 35.

@frostbyte Yes I'll try it in my other PC once the problems get more severe, right now they are random.

@ppd Wish I came across that page earlier one of my Seagate 3TB drives is @ 28% failure rate as per the site, I don't want to lose a third drive in a year that would be demoralising

[insider]

I noticed my hard drives reach temperatures exceeding 40 deg C when I play my game for just 3 minutes, turning on the air conditioner maintains my hard drives at 31 and GPU at 70 deg C, so I hope my GPU lives a little longer, if it dies right now I'll have to buy a 970 or 980, if it lives a few months I can get a GTX 10xx or perhaps get to use an AMD

Kudos to all.

[Zonder]

Well at least it got a few degrees off.

Another thing that I should have though of, is your card over clocked by default? A lot run at higher clock speed by the manufacturer of the card might be worth running at stock nvidia speeds? You will obviously loose a few fps but might help overall.

[insider]

Another thing that I should have though of, is your card over clocked by default? A lot run at higher clock speed by the manufacturer of the card might be worth running at stock nvidia speeds? You will obviously loose a few fps but might help overall.

The GPU is as it was when I purchased it, I didn't overclock it manually and have never overclocked any GPU or CPU in my life, I think it is common sense giving more electricity is going to make it run faster at the expense of heat and wear and tear, I really can't understand why GPU manufacturers themselves provide overclocking tools, the concept is childish really.

Anyway here's my card https://www.asus.com/Graphics-Cards/GTX760DC2OC2GD5/

My game was running at 44FPS it's right now struggling at 30FPS, I don't think I can afford to lose more and it's high time I upgraded, 970,980,1080,Titan,shift to Radeon?, that's going to need a lot of time to assess, my emphasis is on speed but at the same time I can't afford to lose Nvidia technologies like Physx, Cuda, Waveworks, 3D Vision etc.

[Zonder]

Grabbed that from the nvidia site.

GTX 760 GPU Engine Specs:
980Base Clock (MHz)
1033Boost Clock (MHz)

Engine Clock
GPU Base Clock : 1006 MHz
GPU Boost Clock : 1072 MHz

So it is indeed over clocked as standard could try putting it at the nvidia speeds using the asus tool.

[Zonder]

Oh spotted max temp is 97c as well on the site not that I would want something that hot in my system!!

[insider]

Grabbed that from the nvidia site.

GTX 760 GPU Engine Specs:
980Base Clock (MHz)
1033Boost Clock (MHz)

Engine Clock
GPU Base Clock : 1006 MHz
GPU Boost Clock : 1072 MHz

So it is indeed over clocked as standard could try putting it at the nvidia speeds using the asus tool.

Oh spotted max temp is 97c as well on the site not that I would want something that hot in my system!!

Thanks Zonder, I think I'll drop it in my HTPC when the warranty expires which should be precisely in 6 months, 97c I can't imagine that and don't want to either

[mkultra333]

On the plus side, you could put a kettle on top of your card and it'll be perfect for making tea or coffee.

[insider]

[Zonder]

Lowering the clock speeds won't invalidate your warranty as they allow tweaking. The over clock doesn't look a lot but it can be surprising how little can push the temp up and the life of the hardware (goes any over clock that).

I'd have a look at the tool it's straight forward

http://rog.asus.com/4062012/overclockin ... gpu-tweak/

[insider]

Thanks Zonder, was down with a fever for a week so forgot to reply, with the stock settings of a 760 it's 3 degrees lower and monsoon has started here so that gives me an added 3 degrees lower of comfort.

Anyway was just curious there has been liquid cooling for CPUs for as long as more than ten years, I still remember thinking in college days that one day I'll have a liquid cooled CPU, but radiator failures and leakages kept me away from that tech and anyway I have the best CPU air cooler and couldn't be happier, but why has liquid cooling for the GPU not flourished, I remember seeing few GPU liquid cooling sets online but shouldn't that be default now, the maximum a GPU temperature can reach is around 100 C, is that too hot to handle for GPU manufacturers ?

[insider]

Okay I have ordered a GTX 1070 and should reach me within a week utmost, I ran my Ogre project from last year and it takes 99% GPU usage and shoots upto 72C in under a minute in my GTX760, my Unity game also does the same usage and temperature, now the idea to upgrade was because my game is pretty high poly and I was assuming the upgrade would help a lot as 89% of my load is taken by rendering, so after upgrading to a 1070 will the same 99% usage and 72C be still prevalent ?

Average harddrives are meant for 40-45 C, so I am sure 70-75C from a GPU will compromise the entire rig, how do you guys go about it, I can see there are a huge amount of people who play intensive games for hours together, what is the cooling solution you use to the GPU heating issue ?

Or is it yeah you have to live with a 72C GPU while playing games, as a developer it will be 72C, 8 hours a day, for years together what impact is that going to have on my hard drives, on my CPU, on me ?

[insider]

I forgot to post, I finally got my hands on the GTX 1070 and it is huge and quite heavy for a GPU, I played my game for half an hour, temperature never exceeded 55C, GPU Usage @ 60% and 50 million polygons on my screen at a stable 60FPS, now this is more like it

Edit:
And it did shock me when I installed it, the computer wouldn't boot, a little 5 minute googling recommended to clear the BIOS and I did that and it sprang back to life, so if you install a new GPU, just clear the BIOS if your computer doesn't start, there is no need to update the BIOS as recommended on some sites, my ASRock Z87 Extreme4 is still running the 2013 default BIOS and it's great.

And I found atleast a 100 posts saying GTX 1070 doesn't start, clear CMOS one line answer.
Nvidia should add that as a note in their manual, buying a 500$ card only to realize the computer doesn't start after installing it at 12AM isn't a great feeling, I lost my appetite and it gave me a headache