2x MSI GeForce GTX 980 Ti, Idle temp worry

[Drakius]

Okay, so I recently built a machine for myself. Here is what I think are the relevant specs are

i7 4790k
2x MSI GeForce GTX 980 Ti Gaming 6 GB (stock, no manual overclocking, only what came out of the factory)
16Gb RAM
2 monitors which I'm 95% sure run at 60hz
HAF X case with stock fans
Ambient room temp fluctuates between 20C and 25C

Now here's my "issue" I say it that way because I'm uncertain it's an issue. Basically, my cards run idle at around 60C. I also believe that the cards I own MSI made them so the fans don't actually kick in till 60C. So when checking MSI Afterburner, the fans turn on and run at about 30% or less to keep the GPU at 60C

If I make a custom curve I can keep the gpu lower, 50 or mid 40s. Which seem on par with the idle temps of this review. http://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Gaming/34.html

Under full load, my card run anywhere between usually 77C and 83C. I've not seen them go any higher then 83, staying at 83 usually only for a few before dropping back to 79-80 Which seems on par with this review's load temp http://www.hitechlegion.com/reviews/graphics/45141-msi-geforce-gtx-980-ti-gaming-6g-review?showall=&start=6

I feel like the fact that they're SLI, they might run a bit hotter both due to being so close.

That said should I run a custom fan curve to lower the idle temps to 50ish and lower? Or just let the cards do as they intend to do out of the box and keep themselves at 60C at idle? Will that harm the GPU in the long run? I plan on keeping these GPUs for the next few years to come, atleast 4 years is what I'm thinking. I'm kind of assuming/hoping MSI knew what they were doing when they built their card and that if they don't bother to make the fan start before 60C, then anything 60C or lower is not really an issue even in the long run.

Help me feel a bit better and not so worried. I apologize for the novel of text!

 

[getochkn]

More fans in your case and better air flow to get cool air from the front of the case across the gpus and out the back or the side. Also crank up the case fan speeds a bit too. I usually go with custom curves for everything because everyone's case, ambient temps, etc, are always different.

 

[Reaper_7799]

I would run a custom fan curve to get your temps a little lower but you don't have too, the fans can always be replaced much cheaper than the board itself. It probably wont be an issue but that does seem a little high, even with sli. Especially since you aren't even doing anything but it depends on your airflow on your case and ambient room temps also. People with one card have had idle temps in low 50 range so with two, I'd think it could be normal depending on case and ambient temps.

 

[Drakius]

More fans in your case and better air flow to get cool air from the front of the case across the gpus and out the back or the side. Also crank up the case fan speeds a bit too. I usually go with custom curves for everything because everyone's case, ambient temps, etc, are always different.

Why would MSI make it so their card's fan doesn't kick in before 60C? Seems kind of silly to have to run a custom fan curve on something the company built to work that way. As far as more fans. I'm already using all the System fan connectors on my motherboard ( http://ca.msi.com/product/mb/Z97-GAMING-5.html ). Even had to plug one directly to the PSU. So can't really fit anymore in there.

 

[Reaper_7799]

More fans in your case and better air flow to get cool air from the front of the case across the gpus and out the back or the side. Also crank up the case fan speeds a bit too. I usually go with custom curves for everything because everyone's case, ambient temps, etc, are always different.

Why would MSI make it so their card's fan doesn't kick in before 60C? Seems kind of silly to have to run a custom fan curve on something the company built to work that way. As far as more fans. I'm already using all the System fan connectors on my motherboard ( http://ca.msi.com/product/mb/Z97-GAMING-5.html ). Even had to plug one directly to the PSU. So can't really fit anymore in there.

They do it to keep it quiet, so you don't have the fans spinning all the time, maybe for longevity too. In SLI, the non reference cards eject all heat back into the case so even at idle, any heat created will still be pushed right back into your case, so make sure you have enough exhaust too.

 

[Drakius]

I would run a custom fan curve to get your temps a little lower but you don't have too, the fans can always be replaced much cheaper than the board itself. It probably wont be an issue but that does seem a little high, even with sli. Especially since you aren't even doing anything but it depends on your airflow on your case and ambient room temps also. People with one card have had idle temps in low 50 range so with two, I'd think it could be normal depending on case and ambient temps.

I can drop to idle temps of 50C - 45C by running the fans at 50%. 20% more then if I just let the card do it's default thing. Also fan are replaceable yes. But well, just building a PC for me is nerve wracking. Changing a fan on a custom GPU? Yeah I don't really want to do that. Basically I'm trying to balance heat load and fan RPM load to make the cards last 4 years and then build a new system. It's what I did with my SLI GTX 570 build (The one I had before this one).

 

[Drakius]

They do it to keep it quiet, so you don't have the fans spinning all the time, maybe for longevity too. In SLI, the non reference cards eject all heat back into the case so even at idle, any heat created will still be pushed right back into your case, so make sure you have enough exhaust too.

http://i60.tinypic.com/2ekur1g.jpg (Minus the reference card blower out the back) That's the type of airflow I have, intake from the side over the cards, and from the front of the PC, exhaust out the top and the back. But yeah, I think all those factors. Them being so close, cards ejecting the heat in the case, is what makes them idle at 60C with 30% fan. Like I mentioned I can get them to 48ish and under with a custom curve. But that keeps the fan running at 55%. I really want to drag these GPU's life for many years, and well I'm just not comfortable changing fans on a GPU, just building a computer is nerve wracking for me. Hence why I try to make one build last for years before switching it up completely.

So if running the fans at 50%+ to keep an idle temperature of under 50C constantly, will wear the fans out after a couple of years, I don't know if I want to do that.

 

[getochkn]

Maybe put your side fan to exhaust to help pull the heat out that is gathering right around the cards.

 

[Reaper_7799]

You can leave it where it runs, I don't think it will hurt it in the long run. It's mainly the really high temps you want to avoid. Some people just don't like higher idle temps but I don't think it will hurt it.

What I would do though, is change your upper end fan curve to keep your temps below 80, as when you start to get close to 83 like you are hitting it now, could cause some throttling if you don't change your temp curve in afterburner.

 

[corndog1836]

you are in luck!!!! I have a HAF X with two 780ti's in sli overclocked to 1267mhz and had similar issues.. here is how I resolved them!!!!!!

fans installed and orientation++++++++++++

1. 140mm rear back fan in exhaust
2. top rear 200mm fan as exhaust
3. top front 200mm fan as intake
4 . side fan replaced with 200mm led cm fan blowing directly on gpus as intake
5. front bottom fan 230mm as intake
6. modded front drive bays to accommodate 2 140mm led phantek fans in push/ pull configuration blowing directly in front of cpu and top GPU , they fit exactly in the drive bays.

gpu temps dropped 10c ... fans don't need to be spinned very fast either.case never gets warm. top gpu will be 7c hotter during stress tests.


I have a 4790k overclocked to 4.7gz overvolted 1.37v 2 780ti classifieds in sli overvolted to 1.2v and overclocked 1267mhz.. temps on top card are 83c max lower card 75c max in short bursts. mostly stay below 77c at 100 percent use and idle at around 40c.... cpu never gets above 60c(Noctua DH15 cooler) I'm prime 95 . cpu idles 30c

 

[Drakius]

I'll try changing the side fan from from intake to exhaust so it pulls the card's hot air out, like getochkn suggested and report back on what that does to my idle temp.

 

[corndog1836]

I'll try changing the side fan from from intake to exhaust so it pulls the card's hot air out, like getochkn suggested and report back on what that does to my idle temp.

I tried that before and it didn't work as well a flooding the case with cool air... exremly positive intake setup is the key. I am pushing out 1100watts from my psu also. trust me IT WORKS....

 

[corndog1836]

look at my post.. I have to replace the side fam with a better one .... the original doesn't move enough air. my case is a wind tunnel exhausting out the top rear and back ...I have read about this issue for months and my setup works. those aftermarket card have custom fans and heatsinks and NEED to have cool air available for them to suck as well as they exhasust warm air into case that needs to draw out. plus they are all led and look cool,

 

[Drakius]

look at my post.. I have to replace the side fam with a better one .... the original doesn't move enough air. my case is a wind tunnel exhausting out the top rear and back ...I have read about this issue for months and my setup works. those aftermarket card have custom fans and heatsinks and NEED to have cool air available for them to suck as well as they exhasust warm air into case that needs to draw out. plus they are all led and look cool,

Unfortunately I can't really add another fan as my motherboard only supports 3 chassis fan, and I've already got four in there, a fifth for a top intake would be pushing it.

As for getochkn. That didn't really work unfortunately the switching from intake to exhaust. In fact one of the card got way hotter when I did a quick full load test. So I'll return to intake for the side fan.

I might play around with a custom fan curve. Between 60C idle at 30% fan sometimes. And 48C~ at 50% fan. Just to juggle the lifespan and make sure it lasts those 4 years I want it to, though no idea if that would do it. Not constantly mind you, once in a while. Might also see into getting a side fan of more then the current 110CFM that this one is providing me, perhaps if more cool air is blown onto the cards that will help.

Well, seems like that's all there is to it for now. Thank you all for the suggestion and things. My cards have been running fine, and I'm usually at around 80C under full load which is en par with many of the reviews out there and what the card's rated for I think, so maybe I'm worrying for nothing.

 

[corndog1836]

look at my post.. I have to replace the side fam with a better one .... the original doesn't move enough air. my case is a wind tunnel exhausting out the top rear and back ...I have read about this issue for months and my setup works. those aftermarket card have custom fans and heatsinks and NEED to have cool air available for them to suck as well as they exhasust warm air into case that needs to draw out. plus they are all led and look cool,

Unfortunately I can't really add another fan as my motherboard only supports 3 chassis fan, and I've already got four in there, a fifth for a top intake would be pushing it.

As for getochkn. That didn't really work unfortunately the switching from intake to exhaust. In fact one of the card got way hotter when I did a quick full load test. So I'll return to intake for the side fan.

I might play around with a custom fan curve. Between 60C idle at 30% fan sometimes. And 48C~ at 50% fan. Just to juggle the lifespan and make sure it lasts those 4 years I want it to, though no idea if that would do it. Not constantly mind you, once in a while. Might also see into getting a side fan of more then the current 110CFM that this one is providing me, perhaps if more cool air is blown onto the cards that will help.

Well, seems like that's all there is to it for now. Thank you all for the suggestion and things. My cards have been running fine, and I'm usually at around 80C under full load which is en par with many of the reviews out there and what the card's rated for I think, so maybe I'm worrying for nothing.

I have my fans connected with a "y" adapter so two can go to one spot on the motherboard and one directly into the psu running at full speed and its one I bought so its quiet even at full speed....

 

[Drakius]

UPDATE:

So been testing a bit. Made this custom fan curve http://i.imgur.com/QayplAx.jpg

I think I might keep this one. It's basically 1% fan for every 1C. Then it ramps up a bit quicker around 70C and hits 100% at 82C just to stay away from that 83C that might start causing throttling.

As you can see from the picture, idle temp is about 48C at less then 50% fan speed, and that's for the hottest card, one with less breathing room of the two, the other sits at 41C idle with about 40% fan speed.

Full Load tests with this custom curve brought my temps to about 77C at it's maximum for the card with less breathing room, and 71C for the other. All with about 80% fan speed. I think that's fairly adequate. I might keep this custom fan curve.

One without this curve the GPUs run idle at 60C with 30% fan speed. I don't think the extra 20%~ fan speed at idle will really hurt the lifespan of the fans at all.
Two the way the card works makes the fan go from 30% to 0% every couple seconds as it fluctuates between 59C and 60C. Essentially turning on and off constantly and rapidly. I feel like the constantly going from 0 RPM to 800 RPM every few second might actually cause more wear and tear then just constantly running it at a smooth 1000RPM / 50% with any fluctuations being by much smaller and smoother increments.

I /think/ this is the best solution to stop me worrying and get the lengthiest lifetime out of my cards. Might also get a higher CFM side fan to blow more air on the GPUs. Most are fairly cheap, so no harm in trying.

PS: Thank you for all the tips everyone, custom fan curve seem the way to go. Atleast for my situation.

 

[Mousemonkey]

A constant heat if it is well below the 90c limit is not a problem, it's the hot/cold cycling that really kills things so if you are that worried about longevity then don't turn your PC off when not in use.

 

[Drakius]

A constant heat if it is well below the 90c limit is not a problem, it's the hot/cold cycling that really kills things so if you are that worried about longevity then don't turn your PC off when not in use.

So keep it on 24/7 for the next 4 years? (which is how long I want this build to last me at least)

I suppose I could understand if I use it for a while, then turn it off then use it again sporadically. But the way I do it is 95% of the time, I turn it on in the morning. Leave it running all day whether I'm at it or not as I work from home. Then turn it off in the evening before going to bed, which leaves it off for about 8 hours before turning it back on in the morning. Sometime in the upcoming future, I might only turn it on in the evening, so for about 6 hours a day, off for the other 18, depending on personal life.

Basically I don't turn it off and on every hour or so. The power comes on once during the day, then off before I go to sleep. I don't think that would really harm the components lifespan more then just leaving it on 24/7? After a bit of research, the whole debate seems pretty split atleast online about whether it's less stressful or not to keep it on or turn it off at night. So not sure.

 

[corndog1836]

A constant heat if it is well below the 90c limit is not a problem, it's the hot/cold cycling that really kills things so if you are that worried about longevity then don't turn your PC off when not in use.

So keep it on 24/7 for the next 4 years? (which is how long I want this build to last me at least)

I suppose I could understand if I use it for a while, then turn it off then use it again sporadically. But the way I do it is 95% of the time, I turn it on in the morning. Leave it running all day whether I'm at it or not as I work from home. Then turn it off in the evening before going to bed, which leaves it off for about 8 hours before turning it back on in the morning. Sometime in the upcoming future, I might only turn it on in the evening, so for about 6 hours a day, off for the other 18, depending on personal life.

Basically I don't turn it off and on every hour or so. The power comes on once during the day, then off before I go to sleep. I don't think that would really harm the components lifespan more then just leaving it on 24/7? After a bit of research, the whole debate seems pretty split atleast online about whether it's less stressful or not to keep it on or turn it off at night. So not sure.

http://cdn.overclock.net/7/7b/600x450px-LL-7b3765e7_900x600px-LL-adfae5d2_05112011390.jpeg

best setup for HAF X its the same fan config I have...IT WORKS 24/7 on low rpm and its very quiet...

 

[corndog1836]

1. 140mm rear back fan in exhaust
2. top rear 200mm fan as exhaust
3. top front 200mm fan as intake
4 . side fan replaced with 200mm led cm fan blowing directly on gpus as intake
5. front bottom fan 230mm as intake
6. modded front drive bays to accommodate 2 140mm led phantek fans in push/ pull configuration blowing directly in front of cpu and top GPU , they fit exactly in the drive bays.

http://cdn.overclock.net/7/7b/600x450px-LL-7b3765e7_900x600px-LL-adfae5d2_05112011390.jpeg

I just didn't use the gpu fan holder in the pic I linked to you... same thing.. keeps sli gpu's cool at idle and load....

 

[Reaper_7799]

UPDATE:

So been testing a bit. Made this custom fan curve http://i.imgur.com/QayplAx.jpg

I think I might keep this one. It's basically 1% fan for every 1C. Then it ramps up a bit quicker around 70C and hits 100% at 82C just to stay away from that 83C that might start causing throttling.

As you can see from the picture, idle temp is about 48C at less then 50% fan speed, and that's for the hottest card, one with less breathing room of the two, the other sits at 41C idle with about 40% fan speed.

Full Load tests with this custom curve brought my temps to about 77C at it's maximum for the card with less breathing room, and 71C for the other. All with about 80% fan speed. I think that's fairly adequate. I might keep this custom fan curve.

One without this curve the GPUs run idle at 60C with 30% fan speed. I don't think the extra 20%~ fan speed at idle will really hurt the lifespan of the fans at all.
Two the way the card works makes the fan go from 30% to 0% every couple seconds as it fluctuates between 59C and 60C. Essentially turning on and off constantly and rapidly. I feel like the constantly going from 0 RPM to 800 RPM every few second might actually cause more wear and tear then just constantly running it at a smooth 1000RPM / 50% with any fluctuations being by much smaller and smoother increments.

I /think/ this is the best solution to stop me worrying and get the lengthiest lifetime out of my cards. Might also get a higher CFM side fan to blow more air on the GPUs. Most are fairly cheap, so no harm in trying.

PS: Thank you for all the tips everyone, custom fan curve seem the way to go. Atleast for my situation.

Good deal...enjoy. I'm not sure about 24/7 I seem to find split opinions. I've always left mine on 24/7 in the past and have never had issues because I had heard it was less stressful to leave them on, rather than turn it off every night.

I work from home too, so I keep mine on all day long, don't turn it off except if I'm messing with settings or something and then turn it off at night. I may go back to just leaving it on 24/7, I'll have to check into it again.

 

[carlosriosness]

UPDATE:

So been testing a bit. Made this custom fan curve http://i.imgur.com/QayplAx.jpg

I think I might keep this one. It's basically 1% fan for every 1C. Then it ramps up a bit quicker around 70C and hits 100% at 82C just to stay away from that 83C that might start causing throttling.

As you can see from the picture, idle temp is about 48C at less then 50% fan speed, and that's for the hottest card, one with less breathing room of the two, the other sits at 41C idle with about 40% fan speed.

Full Load tests with this custom curve brought my temps to about 77C at it's maximum for the card with less breathing room, and 71C for the other. All with about 80% fan speed. I think that's fairly adequate. I might keep this custom fan curve.

One without this curve the GPUs run idle at 60C with 30% fan speed. I don't think the extra 20%~ fan speed at idle will really hurt the lifespan of the fans at all.
Two the way the card works makes the fan go from 30% to 0% every couple seconds as it fluctuates between 59C and 60C. Essentially turning on and off constantly and rapidly. I feel like the constantly going from 0 RPM to 800 RPM every few second might actually cause more wear and tear then just constantly running it at a smooth 1000RPM / 50% with any fluctuations being by much smaller and smoother increments.

I /think/ this is the best solution to stop me worrying and get the lengthiest lifetime out of my cards. Might also get a higher CFM side fan to blow more air on the GPUs. Most are fairly cheap, so no harm in trying.

PS: Thank you for all the tips everyone, custom fan curve seem the way to go. Atleast for my situation.

i am going to try out your fan curve when i get home. (2) 980ti reference cards.

i also read on another forum that putting the card with the higher asic rating on bottom helped. (something about bottom card working less hard, aka creating less heat for top card).

right now, my higher rated asic card is on top. i will try switching them as well.