Overclock.net banner

Two 1080Ti Cards Bit The Dust While Mining

6K views 45 replies 20 participants last post by  Aganor 
#1 ·
Something to consider given the gold rush to mine cryptocurrency as of late. I already had a couple 1080Ti cards under water for my gaming rig and figured what the heck? Might as well mine while I'm at work and sleeping. I downloaded NiceHash and was making about $15-$20 a day for about a week. My temps were never above 55C and figured everything was running just fine.

Looking back in hindsight, I realized things were happening before the big crash and I think mining might be to blame. First thing I noticed was BF1 was no longer loading. I was getting a DX11 error and this had never happened with these cards. I thought, no big deal. I remember having this issue with this game back when I had my 680. Driver issues, and recent patches were likely to blame I thought. So I rolled back drivers, reinstalled the game, and even re-installed Windows to no avail.

Then 3D Mark's FireStrike stopped working. Wouldn't even load the benchmark once I hit 'run'. Always worked before I started mining. Things are getting weirder.

Then Chrome started acting up. Flickering was out of control. I disabled GPU acceleration and things were fine again. But odd none the less considering everything was fine before I started mining.

Finally things took a massive turn for the worst. I went to stop NiceHash one evening and my computer completely froze. Nothing worked. I couldn't even get to the task manager. So I rebooted. No signal to Display Port. This is bad. I took the Display Port cable and moved it to the other card. Still nothing. I changed cables. Still nothing. Checked my error code on the motherboard and saw the red VGA light. This PC isn't even POSTing. Put my old 680 back in the rig and everything works fine again. BF1, 3D Mark, and Chrome all working normally.

So I'm RMAing both of my 1080TI cards together. What hell this has been. After all of the other issues I've had with this build, and there were plenty, I'm out both cards until I get refurbished ones. All I can point to is the mining. I can't fathom for the life of me what else it could have been.

So what gives? Is mining more dangerous for our hardware than we are lead to believe? I know there are many here who are running their rigs 24/7 for mining purposes, but I'm absolutely terrified to put my refurb cards back to work after all of this. After all, this is $1500 worth of cards we're talking about here. It just can't be good for this hardware to run 24/7 at 100% utilization. I don't think they were designed for this abuse.
 
See less See more
#2 ·
Were you able to monitor your VRM temps on the cards? After continuous loads, if the heat builds up enough, it will overheat and eventually destroy the card. Even with a full water block, you still need great contact -- the end result being that enough heat is extracted in a timely manner that over time, there is no excess buildup.

This is one reason I dread buying secondhand RX 480, 570, and 580's when they come off the mining craze, since they are out of stock everywhere.
 
#3 ·
So I am no pro at diagnostic of cards but funny two cards die?
You would think the odds of mining and having both go would lead me to say mining did not cause your issue.
What is your power supply ?? And are you running two separate cables for each card?
good luck figuring it out
 
#4 ·
Quote:
Originally Posted by cg4200 View Post

So I am no pro at diagnostic of cards but funny two cards die?
You would think the odds of mining and having both go would lead me to say mining did not cause your issue.
What is your power supply ?? And are you running two separate cables for each card?
good luck figuring it out
This. What PSU and what cables used to power them?
 
#5 ·
Both cards at the same time makes it seem very unlikely its the mining.
Silicon wears at different rates, So you should of had one die before the other.

What clock speeds and voltages were you running?
Been mining on my 1080ti for 2 weeks now. First day of mining I noticed back of VRM area was getting very hot so modded the stock back plate to fit on to it which has taken quite a bit of heat away.
Also only running mine at 2ghz on 1.000v which keeps the temps about 38-40c.
 
#6 ·
I've also been mining with 2 1080Tis on the reference cooler no less. So far been going a couple weeks without issue. Temps hover at ~68°C. I've noticed no change in hashrate from ~1500Mhz to 2000Mhz on the core, so I've been running them with +250 on the memory, with power limit at 58%. Just thought I'd mention it.

As for the death of cards, unless there was some funky temperature problem with your waterblock (poor heat transfer from the memory or voltage circuitry) I'd be surprised if mining was the root cause. I'd be more suspicious of power delivery either to or from the PSU. Do you happen to know what your 12V rail settles to under full load?
 
#8 ·
Did you ever try running them one at a time? I get that's a pain because of the water loop but are you sure both are gone?
 
#9 ·
Quote:
Originally Posted by bmgjet View Post

Both cards at the same time makes it seem very unlikely its the mining.
Silicon wears at different rates, So you should of had one die before the other.

What clock speeds and voltages were you running?
Been mining on my 1080ti for 2 weeks now. First day of mining I noticed back of VRM area was getting very hot so modded the stock back plate to fit on to it which has taken quite a bit of heat away.
Also only running mine at 2ghz on 1.000v which keeps the temps about 38-40c.
I thought about the PSU as well. I recently upgraded to a Corsair AX1200i which is supposedly one of the best power supplies out there according to jonnyguru. http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=300

It apparently has Over-Voltage Protection, Over-Current Protection, and Short-Circuit Protection, so I can't understand what could have happened to short out two cards at once.

Is there a way to run a diagnostic to determine if the power supply is faulty somehow? I'm currently using it to power my 680 right now with the same cables and there are no issues. I do understand there is far less power draw from this card compared to two 1080Ti cards though.

I took out one card and the same POST problem ensued. Cards were never hotter than 55C. This was about 10C hotter than just running one 1080Ti. Switched between stock settings and running 85% power while undervolting the RAM by -100 and overclocking the GPU by +100 using MSI Afterburner. I also used seperate cables for each card. One 6-pin and one 8-pin cable for each card.
 
#13 ·
Not a single AMD card has had a problem mining. I have 290X that has mined for 3 years non stop. Have 7970 since 2013 mining. I mined 1080 FE for 2 months with no problem. I personally do no trust NiceHash. I rather run only one load on the GPU and not have it switch.

Voltage is very important. Because most run these card on air we limit power to 60-80% resulting under 0.85v. If you are pushing 2GHz core you are probably 1.0V+ which probably is not the typical load these cards are designed to handle.
 
#14 ·
Did you ever watch VRM temps, or just core?

In the future, you should keep HWInfo open while mining, as you can check for many problems at once:

-You can monitor core VRMs/memory VRMs temps. If they're too hot, they probably aren't properly covered by the water block.

-You can monitor the voltage being fed to the GPU and memory, to make sure they're in a safe range.

-You can ALSO watch the voltage being fed to the VRMs. If your power supply is funky, they won't get a solid 12V which could also cause problems.
 
#15 ·
No VRM temp monitoing on 1080ti,s You have to feel the back of the card or use a IR gun to get a idea.



Did you put the pads on these?
Also that die looks drown in TIM.
And lastly you pulled the plastic off both sides of the pads.
 
#16 ·
Quote:
Originally Posted by brucethemoose View Post

Did you ever watch VRM temps, or just core?

In the future, you should keep HWInfo open while mining, as you can check for many problems at once:

-You can monitor core VRMs/memory VRMs temps. If they're too hot, they probably aren't properly covered by the water block.

-You can monitor the voltage being fed to the GPU and memory, to make sure they're in a safe range.

-You can ALSO watch the voltage being fed to the VRMs. If your power supply is funky, they won't get a solid 12V which could also cause problems.
I do not think FE card should VRM temps. Only core temp and load.
 
#20 ·
Quote:
Originally Posted by bmgjet View Post

But it mines so good lol.
What speed is that 1080 Ti running?
 
#21 ·
Quote:
Originally Posted by bmgjet View Post

No VRM temp monitoing on 1080ti,s You have to feel the back of the card or use a IR gun to get a idea.



Did you put the pads on these?
Also that die looks drown in TIM.
And lastly you pulled the plastic off both sides of the pads.
I did not put pads on the inductors as I understand thermal pads are not needed on those.
I did in fact drown the GPU in TIM. I tend to be very liberal with application as I understand there are no negative consequences to this.
Both sides of the thermal pad plastic were removed.

According to Gamers Nexus these VRM components can take a serious beating (125-150C). If the GPU is averaging 55C under 100% load, could the VRMs really be hitting over 150C with the thermal pads applied?

Since HWInfo does not monitor VRM temps on these cards, is it safe to assume that mining on these is not recommended? I've contemplated getting the 580s, but they're impossible to buy right now.
 
#22 ·
Quote:
Originally Posted by ZealotKi11er View Post

What speed is that 1080 Ti running?
2ghz 1.0v core
5556mhz memory
2.95 - 2.98 HPW efficency.
 
#23 ·
Quote:
Originally Posted by djfunz View Post

I did not put pads on the inductors as I understand thermal pads are not needed on those.
I did in fact drown the GPU in TIM. I tend to be very liberal with application as I understand there are no negative consequences to this.
Both sides of the thermal pad plastic were removed.

According to Gamers Nexus these VRM components can take a serious beating (125-150C). If the GPU is averaging 55C under 100% load, could the VRMs really be hitting over 150C with the thermal pads applied?

Since HWInfo does not monitor VRM temps on these cards, is it safe to assume that mining on these is not recommended? I've contemplated getting the 580s, but they're impossible to buy right now.
I put pads on mine since it had them on the stock cooler.

In the case of tim, the least amount you can use is the best. Direct contact is the best and the more you use the less of that you get.
You only want it to fill in the micro pockets and inperfections on the die and base of the cooler.
Then you also have the issue of it spilling over onto the capacitors around the die, Which messes them up.
Conductive TIM = instant death for those caps.
Non-condictive TIM = does lots of strange stuff.
 
#24 ·
Quote:
Originally Posted by bmgjet View Post

In the case of tim, the least amount you can use is the best. Direct contact is the best and the more you use the less of that you get.
You only want it to fill in the micro pockets and inperfections on the die and base of the cooler.
Then you also have the issue of it spilling over onto the capacitors around the die, Which messes them up.
Conductive TIM = instant death for those caps.
Non-condictive TIM = does lots of strange stuff.
Of course the least amount possible is best, but since GPUs lack a heatspreader, the entire die has to be covered. Too much is always better than too little. This looks fairly similar to a factory TIM application, it's not harmful. If the TIM is non-conductive and non-capacitive then literally nothing can happen to the surrounding components. TIMs based on metal like AS5 could cause issues because, although they might not be conductive, they are capacitive. Modern silicone-based TIMs don't suffer from this issue.
 
#26 ·
There is a such a thing as too much tim. With an open dye like that you want the tim to be applied so it's opaque and you can almost see through it, any more than that and you are actually hindering it's performance because you are creating too large a gap between the dye and heatsink. This can also allow air to get in between the dye and the heatsink, so no when it comes to Tim less is actually better.

Sent from my SAMSUNG-SM-G870A using Tapatalk
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top