Bitcoin Forum

I'm wondering what's the average fan speed some of you use?

I have some GPU's that are ok with 50%, but some need 70%.  I've heard some people say 70% will kill your cards early, while others have said 70% has lasted them years.  Anyone have experience?

I don't think FAN speed has anything to do, it is actually about temperature, running on high temperature for prolonged time is definitely going to reduce life of GPU. However, if fan is spinning at high rate continuously then due to wear and tear, you may need to change fan soon.

So there is no definitive answer whether it lasted long or failed soon, depends upon various factor.

I don't consciously set a global minimum fan speed, but I definitely set a custom fan curve using MSI Afterburner (otherwise I find that the fans on my cards are always unreasonably low at the temps they reach).

Generally my target temp is around high 60s-low 70s. You could probably go slightly higher without significantly decreasing the lifespan of your cards, but it's important to avoid heading too high.

You can also use cheap box fans to help alleviate the strain on the integrated fans. (Just be sure to avoid turbulence wherever possible.)

As Undefined3145 has said I set a custom fan curve to try and keep temps at around 60c.

On my Windows rigs, I set a standard "fan curve" on pretty much everything that goes 100% at 80C, 80% at 70C, 60% at 60C, 40% at 50C, and sometimes 20% at 40C.

 On my LINUX rigs, I generally juggle TDP and fan settings to aim for about 70C card temp when the room is in the mid-80F range, which usually ends up with a fan setting at 80% right now.
 I expect that to drop some as I get my current power infrastructure closer to saturation and have to start pushing "efficiency" harder vs "max hashrate", or start building my "new rig" builds for efficiency.

i run 85% in 2 years..
nothing happen coz not only gpu chip need to cooling down, but also vrm and other component.

Well.. If you are using Claymore miners, you can set your target temperature in your .bat file, and the fans will be spinning to "keep" the temperature steady.
If you are using AB, you can set your custom fan curve and that will be the same.. : )

In my country winter is pretty cold, so in winter time enough 60%. But usualy I set up to 85%

I can only speak for nvidia cards as I own gtx 1060 cards only.
Usually the temp range is from 62 to about 67 with the fans ranging from almost about the same range.

Well, setting up a fixed value for fan speed isn't actually the best idea as they might be running for at an extra speed uselessly.
The best thing would be using a custom curve for the fan speed as that in MSI afterburner.

Since some cards are bad in terms of cooling, the curves differ from one card to another as well (so the sync settings across cards settings is unchecked.)

I guess running the fans at high speeds would only matter with the fans themselves and nothing more.

With my Palit super jetstream 1080ti, the fans needs to be running at 70% just to keep my cards around 65c at 70% TDP for most algos. Don't know why my cards go so hot compared to other cards. I should also say that between my cards it differs 10c for some mysterious reasons.

75 degrees should be fine , I have enough issues with fans going out at that target temp let alone 60 degrees for a target temp.

Although sometimes i wonder if the start stop of the fans using the target temp is bad for them.

In my experience

AMD :

Gigbabyte - terrible fan quality 1 out of every 2 seems to break.
Saphire and MSI - best.

Power color not too bad only one of 80 cards had a broken fan

Asus - bad as well but no where near as bad as gigabyte.

Nvidia

EVGA best
ZOtac worst

the rest are all pretty good.

I run mine at 90% all the time. I'd much rather have to replace some cheap fans on my GPU than the chip itself. Also, it helps disperse heat faster which in turn keeps the surrounding area cooler. However, the profitability of a GPU mining rig will most likely drop off long before there there's any issues with the hardware, at which point you'll probably want to unload your old cards to gamers so you can upgrade to the newer cards with better hashrate/W ratios. So, I wouldn't worry about it too much about it.

if I run windows I set at 75% and try to keep temps under 70


if I run smos   I set at 50%   and set card temps to 69/70   and the software moves fans to what is needed

That sounds good in theory but who wants to buy a card with a broken fan, and its not that easy to replace some fans , alot of the newer cards are custom deals , not any random fan works with the screws and connections.  IMO any card whose fan dies in less than six months
to a year even in constant use need to be avoided.

All i can say is avoid Gigabyte (lower end ones RX and Asus)

Keep your cards below 70C, do this with --limittemp=70, that way all software quits

Have little 10cm fans in front of the cards, one per every two graphic cards, most rigs have a slot for fans

have a big fan 50 cm that blows cool air perpendicular below the cards to keep power-supply and cpu cool and bring cool air in below of GPU line-up.

Set the power 60%, say you have a gtx-1070, then the power is say 150W death, I run at 110Watts, and they run full clock at 59C, never come close to the temp-limit.

For 1060, run at 95 watt per "nvidia-settings -pl 95 -i x', where x is gpu number

Now to fan's "LEAVE THEM ALONE", default to AUTO, having the fans run at full just makes noise

The way you keep your gpus cool is with the little fans in front ( facing the connectors ), and the big fan on the end of rack blowing cool air below and over the top of  the cards.

LEAVE the fan alone, all you do is burn out the fan motor on your GPU card running at 100% and then you have a DEAD-CARD.

Little 12v fans with led lights costs $3 each, the 50cm big 110/220v fans cost $10, one per rig, and little fan 4 per rig, so cooling cost is $22 per rig.

It's important to keep the GPU cool for long-life, and to MINIMIZE the money you waste on POWER,

The little fan on the GPU card is designed for a solo card in a box, its NOT designed to be surrounded by 1/2 a dozen heaters.

I did not OC my GPU, so I just put it as automatic, no changes to fan speed.

GPUs are not created equal either because of manufacturer, model, version number, thermal paste application, etc. But even if you already have GPUs of the same exact specifications, there are still chances that their temperatures will vary. So fan speed will really have to vary per GPU.

What works for me, in my local ambient temperature, is a minimum fan speed of 60%. I suggest you gradually check what temp your cards are running with different fan speeds. You will really have to do this yourself and not rely on other's fan speed.

I use 60% for my nvidia cards, the temps are under 65C.
For AMD VEGA, auto fan speed works well, I just set target  temp, always 2200rpm @55C, 1600rpm@65C.

I set mine to auto.
But temp is 72C and fan speed 30%.

Using trixx.
Strange

Very wide variation depending on Ball Bearing or not, and on the specific card.

I've had a couple Sapphire cards with very small fans that need 80% to say reasonably cool run at 70-80% for 4+ years without issues.
I've had Gigabyte 3-fan Windforce models die in a year or less with 60% fan.


I generally aim for cards to stay in the 65C range on my AIR CONDITIONED workbench, so that when they go in the "production" room that runs a lot warmer in the summertime they'll stay in the 75C or less range - and the fan speed ends up wherever it needs to be.

Riser rigs help a LOT on that, much better airflow and less heat hangs around so you can set your fans 20% lower in many cases while seeing LOWER temps.

wow 60-70°C for a GPU, seems very hot. Has anyone ran such temperature for more than a year? I keep my GPUs around 50°C (or even lower in some cases) with a fan set to 50%.

The minimum speed of rotation of coolers on my Asus 1060 6gb video cards is 10% in winter (at an ambient temperature of 0 degrees) and 50-60% in summer (at an ambient temperature of 30 degrees). Cards after Undervolting.The temperature of the cards themselves does not exceed 65 degrees.

been running my fans at 70% - cards stay around 50-55 (rx560's)

I'm using TDP 65%, fan auto and my Gigabyte Xtreme 1080ti water cooled runs at 39C. Evga branded gpus FTW3 goes between 60C-63C with auto setting.

Better to target temperature rather than fan %. Under 70°C is recommended, personally I have all my cards target 66°C which is easy to achieve in most cases.

Depending how your cards sit some fans will run at 35% some at 80%, but fans are cheaper to replace than silicon, so don't worry if they are at a higher percentage on your rig.

For AMD I use  OverdriveNTool (https://forums.guru3d.com/threads/overdriventool-tool-for-amd-gpus.416116/) to set a target temperature.

I've had many GPUs survive 4+ years of high-load usage running in the 70-75C range.
Targeting 50C is WAY excessive, and the only cases of 50C GPU temps with 50% or less fan I've ever had were on gear in an A/C room on risers, or the "outside" card in a "no case" rig with all cards on the MB, or a rig with ONE card in it - and the card was set to run at "efficient" settings.

Water cooled/hybrid is another story, water cooling the GPU is a lot better at keeping the temp down - at a significant cost for the higher-end cooling solution.

Were those AMD or Nvidia cards? I wanted to be on the safe side with 50°C but as it doesn't affect performance, it can't be a bad thing.

All of my 4+ year old cards are AMD, the oldest NVidia cards I have are "only" about 2 years old and spent a lot of their lifetime Folding (which tends to be HARDER on a cards than most cryptocoin algorithms are).

I run 60% on most everything, up to 70% on really hot cards. If you need more than that I think another solution is needed like looking at spacing/ambient.

For me auto fans , gpu target temp <75 C , all good

I found this topic useless.
You need to be specific!  >:(

I do have Vega 64 cards and the min is 3000RPM because the problem of this card is HBM temp and not CORE temp. The rpm is not related to HBM temp, so, you'll need to keep the min high since memory temp will be higher than core always.

Vega card is most often used in cryptonight algo that do not generate a lot of heat compared to others, i think longevity wont be a problem.

OverdriveNTool, set gpu target 57C

i like to keep my GPUs cool, around 69c. That means around 70-75% fan speed. My oldest card is a 1070 that i bought the day they were released, and it is still going strong. So im sticking with these setting becourse they seem to work for the Nvidia cards.  :)

I honestly just mod the GPU temps with afterburner and leave fans at auto. (Although I heard there are much more efficient ways to mod other than afterburner, I prefer it's GUI). Main reason why I do that is because on most of the GPUs I run the fans are the first thing to go.

well about setting the fan on 70% will kill your gpu but settings for my cards depends on temperature of the card itself so if card temp were 60 the fan 60 and so on sometimes i make below that with 10 degree in the cold days
and remember the hot temp of a card could also kill it

I prefer temp ~65c. For some cards set 50%, another 65%, but need see not only for %, chech fan speed it can be 1200 ( 40% ) on 1 card and 1600 ( 40% ) for other card if they have don`t same cooling  ;)

Dont run gpu coolers on high speed, more than 50% this kill fans very fast.
Use extended 12mm cooler for better cooling, this also will be silent

I set mine to a low temperature and power threshold.  The system determines the minimum fan speed based on GPU Temp, which doesn't get too hot because the power is scaled down.

I've heard many different opinions and the last was that you must keep the fans at 100% and this will extend the life span of the gpu's. I run my fans 70% and the temp is around 82c and you could not touch the cards.They're extremely hot. But i think my problem is not from fan adjustment instead of space ventilation. I have placed too many gpu's in one room and have no option to put some king of fan to get our the air. Do you think this temp can cause some damage to the cards - evga 1070 ti.
Cheers!

Problem is the fans on some GPUs break pretty easily if you keep em at 100% for a long time, and replacements cost time and money.  


At the very least 80c+ is subpar and you should consider moving some GPUs to another place to lower temps or undervolt to reduce power consumption and thus lower temps a bit. My oldest RX 470 ran at 80c for a year and it's still going fine, though I moved to a cooler spot a few months ago.

I did't play with Temp/Fan settings and my cards with modded BIOS runs at 65-69c with a fan speed around 44%.

I run my fans at 80% to keep cards cool and gain more hashes. My cards run 38-53 degrees. 1080ti`s

Most fans used in GPUs don't handle extended time at 100% - the BALL BEARING ones might last a while that way, but far too many GPU makers have moved to cheap sleeve bearing fans that won't last 6 months at 100% and might not last 6 months even at 80%.

Wear is NOT linear with speed, it's more of an exponential function - that last 20% generates MORE wear on the bearings AND SEALS than the first 80% in most cases.
Also, that last 20% generally only adds more like 10% or so to the airflow, normally not worth the wasted power much less the added wear.

CASE fans, especially non-PWM types and anything Rotron/Delta/YS Tech/NMB (couple more I'm forgetting at the moment) ARE normally designed to handle 100% fan speed for extended periods of time.

I have Asus GeForce GTX 1080 Ti Turbo GPU's, with auto-fan mining temperature is between 80-84°C, which looks way too high to me.
Putting fans to 100% gives temperature between 49-62°C, but I use NvEZOS as OS (ewbf miner) and don't see any option to lower the temperature with auto-fan (I guess that something about 70°C would be optimal).
Any thoughts?
In NvEZOS OS there is also Claymore miner, should I use it instead of ewbf?

I do not know who, what he says, but I think that fans are consumables, so you do not need to save cards.
Accordingly, everything is simple - we always put 100% cooler on video cards and forget about those thoughts!
Replace the cooler from aliexpress is very inexpensive and fast, in case of emergency you can fasten the usual 92x92 or 120x120 cooler on the cards, while waiting for the normal with ali. And if you are lucky and your coolers can be disassembled and not broken, then generally there are no problems, greased and drove on!

For people with only a few graphics cards (especially if they are not the same) it's not convenient to dive into a risky procedure of changing the fans. Also, they must have a stock of fans or wait a few weeks for Aliexpress to deliver new fans. They will loose too much time and money waiting. Personally, I prefer to stay on the safe side and keep everything running without problem as much is possible, even if I have to sacrifice a few percents of hash rate.

So ordering some spare fans for "just in case" would be an good option?
As I can see from Aliexpress, those are from 5-20€/piece.

More power consumption. You can normally use 50%-80% fan when mining 24/7. It depends on your cooling setup as you don't need 100% if your ventilation is properly planned. Remember the GPU fan is designed for being in a closed case and most miners are using open air cases.

Fan speed depends on which room is the farm, closed housing or open, if you have fresh air, air conditioning, the distance between the graphics cards, but in any case from 70 to 90% is acceptable, in this mode if you do Maintenance every 6 months, the fans of video cards long live.

Most of the cards in my riser rigs run at 60% OR LESS fan.
I think the lowest one right now is at 40%, but I'm still in the process of rebuilding my old "folding" rigs into riser rigs.