RTX3080 core clock drop when mining

[Etherion]

Hi. I noticed that my RTX 3080's drop there core clock. Any idea why this is?

GeForce RTX 3080, 885 MHz, 9501 MHz, 46, 205.02 W
GeForce RTX 3080, 885 MHz, 9501 MHz, 49, 208.43 W
GeForce RTX 3080, 855 MHz, 9501 MHz, 41, 196.36 W
GeForce RTX 3080, 870 MHz, 9501 MHz, 41, 195.91 W

I have changed the powerlimit to 340 and got no oc set.

THis is on ubunut 16.04 NVIDIA-SMI 455.23.04

[sakmsb]

Hi. I noticed that my RTX 3080's drop there core clock. Any idea why this is?

GeForce RTX 3080, 885 MHz, 9501 MHz, 46, 205.02 W
GeForce RTX 3080, 885 MHz, 9501 MHz, 49, 208.43 W
GeForce RTX 3080, 855 MHz, 9501 MHz, 41, 196.36 W
GeForce RTX 3080, 870 MHz, 9501 MHz, 41, 195.91 W

I have changed the powerlimit to 340 and got no oc set.

THis is on ubunut 16.04 NVIDIA-SMI 455.23.04

Known issues with almost all 3080s.

Bad thermal pads is one of the issues, could be other issues as well.

[Etherion]

name, clocks.current.sm [MHz], clocks.current.memory [MHz], temperature.gpu, power.draw [W], clocks_throttle_reasons.sw_thermal_slowdown
GeForce RTX 3080, 825 MHz, 9251 MHz, 49, 196.94 W, Active
GeForce RTX 3080, 900 MHz, 9251 MHz, 50, 210.01 W, Active
GeForce RTX 3080, 825 MHz, 9251 MHz, 43, 191.87 W, Active
GeForce RTX 3080, 825 MHz, 9251 MHz, 44, 189.88 W, Active
Gigabyte GeForce RTX 3080 Egal OC, Cant handle the heat it seams.  clocks_throttle_reasons.sw_thermal_slowdown = Active
SW Thermal Slowdown
This is an indicator of one or more of the following:

Current GPU temperature above the GPU Max Operating Temperature

Current memory temperature above the Memory Max Operating Temperature

Darn going to RMA then. Can't use them as is

[jgonzi]

name, clocks.current.sm [MHz], clocks.current.memory [MHz], temperature.gpu, power.draw [W], clocks_throttle_reasons.sw_thermal_slowdown
GeForce RTX 3080, 825 MHz, 9251 MHz, 49, 196.94 W, Active
GeForce RTX 3080, 900 MHz, 9251 MHz, 50, 210.01 W, Active
GeForce RTX 3080, 825 MHz, 9251 MHz, 43, 191.87 W, Active
GeForce RTX 3080, 825 MHz, 9251 MHz, 44, 189.88 W, Active
Gigabyte GeForce RTX 3080 Egal OC, Cant handle the heat it seams.  clocks_throttle_reasons.sw_thermal_slowdown = Active
SW Thermal Slowdown
This is an indicator of one or more of the following:

Current GPU temperature above the GPU Max Operating Temperature

Current memory temperature above the Memory Max Operating Temperature

Darn going to RMA then. Can't use them as is

Make a simple trick and you Will see is not a RMA Problem… put MEM -1000Mhz... I have Gigabyte too and the problem is the Thermal Pads are very poor.. you need to change them to be able to mine to +100Mhs...



I add a picture mining with lolMiner with the Gigabyte 3080...
 

[sakmsb]

Make a simple trick and you Will see is not a RMA Problem… put MEM -1000Mhz... I have Gigabyte too and the problem is the Thermal Pads are very poor.. you need to change them to be able to mine to +100Mhs...



I add a picture mining with lolMiner with the Gigabyte 3080...
 

Didn't understand what you meant by put MEM -1000 Mhz and how you are managing 100 MH/S, Please elaborate.

[jgonzi]

Make a simple trick and you Will see is not a RMA Problem… put MEM -1000Mhz... I have Gigabyte too and the problem is the Thermal Pads are very poor.. you need to change them to be able to mine to +100Mhs...



I add a picture mining with lolMiner with the Gigabyte 3080...
 

Didn't understand what you meant by put MEM -1000 Mhz and how you are managing 100 MH/S, Please elaborate.

The -1000Mhz was only to show you that you don't have a RMA, you have temp problems in the memory... and you will have it also with a new card, and also with another brands...

With the -1000Mhz the memory get less hot and i will be stable from the begining... of course you will get only +70Mhs but they will be stable... because there is memory trolling.

My photo is after the solution of the mem temps...

[rdluffy]

The -1000Mhz was only to show you that you don't have a RMA, you have temp problems in the memory... and you will have it also with a new card, and also with another brands...

With the -1000Mhz the memory get less hot and i will be stable from the begining... of course you will get only +70Mhs but they will be stable... because there is memory trolling.

My photo is after the solution of the mem temps...

I have to disagree  
If you have to put -1000Mhz, or change the thermal pads because it's horrible, yes, it's a good reason to RMA
After you changed the thermal pads, your card is running 24h stable at 100mhs?

70 Mhs consuming 235w don't look so good, the good thing about RTX 3080 is the 100Mhs for 250w  

Nvidia said it's all good now, but for me it's clearly that these cards have a problem, drivers, capacitors, I don't know, but there's something wrong

[reb0rn21]

I have core drop with ubuntu 20.04 and gtx 1070... as soon it pass 54C is drop 10%... with ubuntu 16.04 had no such issue

I am not sure if 18.04 can support them or they added some same crap in newer NV drivers

but if you are on 16.04 then defo some protection to core drop is added in driver!

might test in windows if same shit happen

[sakmsb]

My GPU temps are as low as 32C still is throttling and core cannot stay above 1000MHZ.

Contacted Gigabyte support and they are of no help and asked to contact the reseller, even reseller is not willing to help.

I am basically out of options and may be will have deal with changing the thermal pads myself, which I'm really not confident about.

[Metroid]

My GPU temps are as low as 32C still is throttling and core cannot stay above 1000MHZ.

Contacted Gigabyte support and they are of no help and asked to contact the reseller, even reseller is not willing to help.

I am basically out of options and may be will have deal with changing the thermal pads myself, which I'm really not confident about.

You could lose a 1000 usd gpu ehhe, you have to be very careful, mining 24/7, that card won't last 2 years the way things are. Ask them if you can change thermal pads and still have warranty with them. It happened to me once and xfx accepted.

[jgonzi]

The -1000Mhz was only to show you that you don't have a RMA, you have temp problems in the memory... and you will have it also with a new card, and also with another brands...

With the -1000Mhz the memory get less hot and i will be stable from the begining... of course you will get only +70Mhs but they will be stable... because there is memory trolling.

My photo is after the solution of the mem temps...

I have to disagree  
If you have to put -1000Mhz, or change the thermal pads because it's horrible, yes, it's a good reason to RMA
After you changed the thermal pads, your card is running 24h stable at 100mhs?

70 Mhs consuming 235w don't look so good, the good thing about RTX 3080 is the 100Mhs for 250w  

Nvidia said it's all good now, but for me it's clearly that these cards have a problem, drivers, capacitors, I don't know, but there's something wrong

I think I didn't explain it properly...

It is not a card problem, it is a temp problem for eth mining., the cards works perfectly for gaming...
If he make a RMA:
- 1st he will have to wait a lo for another 3080.
- 2nd he will have the same problem in the new one, also other brands have that problem, for example some MSI...

The solution accept that the thermal pads are bad quality:
- reduce mem - 1000Mhz and also PL to 180W to have stable 70+Mhs
Or
- change thermal pads and mine 215W at 96+Mhs or 240W to pass the 100Mhs( I prefer 215W - 96Mhs)