Bitcoin Forum

*While in Phoenix miner I noticed that pcap power changes to pcap temp and hashrate drops*


My Gigabyte 3080 Eagle OC drops hashing after mining for a minute or two.

Tried changing risers and power connectors to no avail.

I m on windows 10 with latest Nvidia drivers. I have Silverstone Platinum PSU 1200 W. The hashrate drops on both everything stock and TDP at 80 or 70 or any other value as well.

Please help!

It's only with Phoenix miner?
What about claymore?

There's the problem of popcaps, but if you have newest drivers, and also with reduced TDP, its hard to blame popcaps
Looks like a app problem, I think all miners are not yet optmized to run new nvidia cards, but try claymore for some time and tell us if this problem persists

Your hardware is good, PSU, all system updates...

Memory is trottling, it is very hot on 3080, 100+ degres.

Same hashrate drop with claymore as well.

I have not tried other algos yet as I recall KawPow will use more power and definitely cause the temps to rise even more


I guess Alexgtr is right, temps in MSI afterburner shows 50 to 52 degrees Celsius but internal temps seems to be off.

Yes, KawPow uses a little more power
I have a card here that is very hot at the memories, it's always on 92 to 96 degrees but the GPU is 55c

But this is easy to find, I suppose you are using MSI afterburner, or another app, just put the fans on manual, and set something like 60 or 70% and see if the card stop to trottling, you can even put 90% just to test
Maybe the fan curve is not enough to keep the card running 100%, and above certain temp the card will reduce power to reduce temp

In MSI afterburner, I have set the fan at 100% manually to test it but same results. The hashrate drops to 70-85 MH/s fluctuating frequently.

Yes the power consumption also drops when hashrate drops along with change of pcap power to pcap temps.

I removed the card from rig frame set it aside and I have just put a table fan at high speed near the card and have cold/fresh air blowing on it. Looks stable for an hour+ now. But I guess this is not the solution. ~97 MH/s now.

https://imgur.com/a/fVKJebW

https://i.imgur.com/UlN4MNt.jpg

I purchased it from a local online retailer, and they do not have any more stock to replace the card it seems.

It's weird, if the card is hashing normally now you can conclude that is a temperature problem
If you already tested with fans at 100%, another thing that could be are the thermal pads, can you check if thermals are in touch with heat pipe?

Or maybe the position of the card in your rig, is there hot air blowing direct to this card?

I have never disassembled a GPU, lil afraid to do so. I have been seeing some videos on youtube.

Does GPU also have a thermal paste similar to CPU, If I decide to; can I change the paste and see if it helps?

I had this card in an 8 GPU set up the usual way we set up rigs on a mining frame, spacing between cards is enough though to avoid hot air from the adjacent GPUs to directly blow on this.

Shall I get a GPU cooler fan something like this might help https://www.amazon.com/gp/product/B07XCFB5WP/

Weather wise it is getting cold here in Arabia so I will not run into heating issues until March with this type of table fan set up, but after that.....

Do not disassembly your card because you'll lose warranty
Wait a little more because more people can help you to find a solution here, it's not normal to need a table fan to keep your card running 100%, and you already reduced the TDP, so the card is not running at full power or above specifications, Nvidia are good to lock the bios so it's hard to mess with the card

3080s havent been out long enough to have worn out paste. 

There was a lot written about the capacitors on back side and Gigabyte used the worst so that may be the issue.  It could also be just bad paste from factory.  Thus good advice not to disassemble as you can always warranty the card.

Not all gpus are equal.

Change the thermal pad for a good ones and the problem is solved. I had that problem too... You can see how it changes. The disipator is good the thermal pad no... I hope you can see the image

https://ibb.co/92fH9ty

And then how to apply under the guarantee?
If you disassemble a video card to replace thermal pads, the manufacturer or seller may refuse to provide warranty service. It works like this in my country. Even if there are no warranty seals on the bolts of the video card (as is usually the case with Gigabytes), the opening marks are visible under the microscope.

So the history repeats again, my Gigabyte memory is hot as hell  :D
But I managed to fix with fans at 60%

The worst part is you can't open the card and stay under warranty

In your case, the thermal pads are this bad or they didn't touch the heat pipe?

can you post memory temperature of 3080 when mining ETH? As far as i know, GPU-Z and other popular monitor tool can not read memory temperature of 3080.

Igorlab post a pic showing 104oC under load, but not mining. The name of the software also hidden.

Complete true... new FE comes with the thermal pad... If you don't do that mod of thermal you will destroy the memory... So what is better... guarantee or destroy card...

Ok maybe the problem of 3080 FE version, but what about AIB like MSI trio, Asus TUF, giga aorus...?

There are other forum with people informing that hashrate goes down... in TUF and MSI too the problem is that we cannot see the memory temp... Igorslab hat a Nvidia Program to see it... but we are not lucky to have it.

You can see you have a memory temp problem if you leave the fans in auto and the fans goes up and up when GPU stills cool. 

In windows if you use GPUZ you should put programs that make the Memory Control Load high... 90-100% miners are perfect. if you don't want to make the thermal mod and you have this problem, there are other solution direct cool to the card or reduce memory Mhz for example-1000Mhz. With that reduce you can also limit the power limit to 180W and you will get a fix 77Mhs... and fans want be at 100%.

We also don't know if the spin fan to 100% on memory control is done in all brands. In gigabyte we know that his BIOS check the memory temps, which is great... what is not great are the quality thermal pads.

If some opened it, he will also see that backplate come with the exact place to put thermal pads of 3mm. So they have that prepared to do it to help to reduce temp but they didn't put them...
For example I have made a stress test in a room of 35C... having 100Mhs+ fans in auto ... the hash rate was completely stable and fans where in auto at 72% and GPU Temp 63C... the memory I don't know but surely not near the 104C Trolling Memory Temp

Thank you for your info. Can you share the link that other forums discussing about 3080 hashrate drop? I want to read more about this issue.

https://miningclub.info/threads/vse-problemy-rtx-3080.70379/ (https://miningclub.info/threads/vse-problemy-rtx-3080.70379/)
https://www.igorslab.de/en/nvidia-copies-the-pad-mod-from-igorslab-for-the-geforce-rtx-3080-fe-to-mimic-the-hotspot-for-the-gddr6x/ (https://www.igorslab.de/en/nvidia-copies-the-pad-mod-from-igorslab-for-the-geforce-rtx-3080-fe-to-mimic-the-hotspot-for-the-gddr6x/)

If you read the original post of the thermal pad mod on the FE, you will read that miners had apologies about that problem with the FE too.. You can also find some people problems in Reddit.com... what it is true that for gaming you Will not have problems… the games don't make so high use of memory control

So this table fan set up is working stable. For a few hours during the day when temps are high outside the pcap changes to pcap temp and during late evening or so it reverts to pcap power.

The good this is the hashrate drops only ~2 MH/s now as compared to 20+ previously.

I am planning on ordering this Graphics Card Cooler, ~2000 RPM which should be hopefully good enough.

https://www.amazon.com/gp/product/B07XCFB5WP/ref=ox_sc_act_title_2?smid=A235LT0EDLFSAR&psc=1

I think you should sell that card and buy another one. TUF, Strix, Trio, EVGA... should be fine

TUF have reports problems too… and some MSI too… so it is normal

I think we need to wait for the miner developers to update them. Use desktop fans. i do just that

Since today morning the card has dropped hashing to 70 - 80 on the same settings, no matter what I do.

Within few seconds it drops from 100 to 70 - 80.

Pathetic experience with 3080

I don't know what you can do anymore, you tried everything already and still suffering this problem
Only a few people have these new cards, and it's hard to gather more information, there's only a few people mining with these cards, and I don't know when devs will have the chance to get one in hands to improve the miners (Claymore, Phoenix etc)

If there's plenty cards in the market, at this moment people would be discussing more

You have only two more options, wait until update of miner, like the user above said or try RMA your card

sell it while you have the time then you buy the new ones that will come later on or rma it, clearly your gpu is at faulty here.

I would RMA its not worth the $700.  I just got an evga and its 91-101 no problems.

 $1100 in middle east.

Have contacted Gigabyte support, waiting for their response.

it would be useful to list all brands (and model) without temp problems. EVGA model?

that mean maybe we need to skip 3080 first lot for mining?


3070 have this problem or not?

The fact is nobody knows this answer yet
There's a problem about capacitors: https://www.igorslab.de/en/what-real-what-can-be-investigative-within-the-crashes-and-instabilities-of-the-force-rtx-3080-andrtx-3090
Manufacturers say there's no problem, while people had a lot of problems, and Nvidia released a new driver to solve this issue, but reducing some configs

And like I said above, there's no GPUs to buy and people are not testing and there's no way to figure out what's happening, at least for now
If you don't have money to waste, maybe it's better to wait until new batch of cards, if you have money and want to buy, look for the "premium" models of cards from Asus or EVGA for example, avoid the simple ones

The capacitatora were a driver problemor an excess of OC... but we don't normally see this problems on Nvidia only on AMD...

I have the same issue as you, with an RTX 3080 Gigabyte Eagle OC. Starts out at 88Mh, slowly dropping to 77Mh.
I installed the GPU straight on the motherboard since this is also my gaming rig. I noticed that the reason for this is that the core clock is dropping automatically to around 1000Mhz for some reason.
Also, total power consumption according to Phoenix miner is around 200W. (same thing with Claymore btw)

Did you find a solution for this problem? Thanks

After a few days of mining the clock drops further low to 800 and even lower. Just pathetic cooling.

Some say replacing the thermal pads worked for them but this will void warranty.

https://bitcointalk.org/index.php?topic=5287329.msg55545084#msg55545084


Here's a thread about similar issues with other 3080s.

https://miningclub.info/threads/vse-problemy-rtx-3080.70379/post-1840603

see that is the problem when they do a shit job, imagine if you tell them that you are returning the card because thermal pads are bad, they will never let you rma it and if you try to fix the shit job then you are the one to lose in the end, my point is, warranty law should be revised, reason i think in the future, you will buy only the gpu core chip and place inside the gpu backbone that you already have, just like your cpu. You will buy the gpu board memory and core and that is how supposed to be, right now any silly thing happens to the gpu, you need to send back.

Yes, definitely they need to either sell good stuff or revise warranty policies. For someone outside the U.S is a real pain for RMA.

thanks...my russian is a little but rusty  ;D

I put that solution... the disipator is well made, but they put shit thermal pads... probably 3 or 4$ cheaper... for gaming you don't need them but for mining with that thermal is not enough and you need some with minimal 8W/mk

LOL, Card is back at hashing 100 MH/S as the ambiance temps are going down here.

https://i.ibb.co/vxRfHZG/Inked3080-LI.jpg (https://ibb.co/VW5bt2Z)

last lot of 3080 have hash rate drop or any problem?

Always this problem is Memory TEMP... reduce Memory temp and no problem

Did you do anything to hash 100Mhs again?
New drivers, less ambient temperature?

I almost got a 3060ti but I'm trying to get a 3080 and found this exact model, 3080 eagle, but I remembered this thread and came to see if it's worth or not, it's the only model available right now

With a Gigabyte 3080 too 100Mhs in lolMiner:

https://i.ibb.co/YywCbPt/Nvidia-Lolminer.jpg (https://imgbb.com/)

You need to make the Thermal Mod or add a really good ventilation. But with a replace of the Thermal Pads memory works great.

without dissambling and see how memory cooled hard to say.some vendors do very bad memory cooling, some memory on air only half memory cooled like that. besides you can dissamble it without warranty lost but need to be very careful here on youtube but it is russian but you can see how you make it :)

https://www.youtube.com/watch?v=82sf5Lk2lqA&t=1301s

hot gun the carefully take sticker.

With Gigabyte it doesn't come warranty stick... that's good to make the thermal change

I didn't do anything new. But I'm considering adding extra thermal pads to cool the memory.

Like Game of Thrones "Winter Is Coming" here. Ambiance temps is in single digits during night and it hashes 100MH/S only during night time.

I still have that table fan set up blowing air directly on GPU fans.

He added sensors on memory to see the temperature, nvidia doesn't, he does, smart guy. I think his card is msi.

There is this video of red mining panda on a 3080 gigabyte, he added thermal pads on the back and it worked without any problems, I guess thermal pads only on the front is not good for gddrx6.

https://www.youtube.com/watch?v=cDau0RyfJ0I

If red mining panda had put a sensor before those thermal pads then we could see the temperature the gpu was throttling and the temperature after he had placed the thermal pads.