EVGA RTX 3090 XC3 power limit and low hashrate only in hiveOs

[eirikmars]

I have an issue with my EVGA RTX 3090 XC3. My 3090 is powerlimited in hiveOs for some reason. I wont get past 110mhs(+1000, everything above wont affect the card) at around 280w(I didin’t touch pl in hiveOs, I set it to stock).

I connected the card to windows to check if the memory junction temps are too high, but it isn’t and I had the same issue in windows; that it wont use more than 280/290w. What fixed the issue for me in Windows was increasing the pl to 104%, then it mined normal like all my other 3090 cards and I can set memory oc to +1500 in afterburner no issues at around 305w. But I don’t want to run the card in my main computer, so do you guys know a fix or a tweak I can do in hiveOs to fix my issue?

I have tried setting the pl in hiveOs to 366(max value for my card), but same result. It is stuck at 280w and “only” 110mhs

[eirikmars]

[REIFJGH]

I got the same problem. I was wondering if someone let me know the solution.
Thanks.

[ne0pol]

Hi @All,
first, I have the same problem myself.
My guess is that the temperature of the RTX 3090’s memory (GPU Memory Junction Temperature) is too high and it therefore throttles.
I have two identical RTX 3090’s. One also goes up to 280-285W and the other goes up to 290W.
It looks like something is throttling / protecting the graphics cards. I have no other explanation

Unfortunately, we cannot read the “GPU Memory Junction Temperatureout” at the moment, but hope dies last. (see Link)
Link:

Request: GPU Memory Junction Temperature via nvidia-smi or NVML API

BR

[colo]

hola yo la tengo con un oc de -180 2000 300 y anda en 118 m/h

[eirikmars]

Thanks for the respond, but I checked in windows and it isn’t. Runs stable at 126 in windows.

[Dcaron9999]

Same problem here. Recently added three 3090 xc3 ultra cards to my hiveos 0.6-210 rig. 2 cards refuse to go OVER 265 Watts and around 100MH/S. Moved one of the suspect cards to Windows 10 gaming PC and used afterburner to move power limit to 102% and had 298 watts with 118MH/S…

My conclusion is that the drivers in Windows are more robust and up to date than those in Hiveos…

Anyone find a workaround in hiveos?

[edwinlin1012]

Hello ,

I have the same problem as you. I also have 3090 cx3.

I tried Core 1200 and Mem 1600 pw 300 to get 117MH/s,

It is the maximum limit value in hive os, and I can get 120+mh/s in windows.

[Rscupra]

Have the same problem, anyone got a solution in Hiveos?

[FrankieMiners]

I used to have the same issue… power limitation

Yesterday I’ve installed drivers 470.82.00, then (I don’t know if this is something related to the new drivers) this option appeared

I activated with the following OC
FAN 90%
CORE 1100
MEM 2350
PL 360

And the 4 3090 (GeForce RTX 3090 24268 MB · EVGA Micron GDDR6X · 94.02.42.C0.10 · PL 100 W, 350 W, 366 W) started to mine at 121.4

I was unstable, so I unchecked FORCE P0 STATE and everything worked like a charm!

To be honest I don’t know if the reason that solved the issue was the new drivers, the FORCE P0 STATE option or the 360 PL OC… but it’s working!!!

[alkilo]

Hi guys,

I had the same issue and i found the solution.

First of all great thanks to FrankieMiners because you gave me the solution.

I see you are in Nvidia Drivers ver : 470.82.00, and i was in version : 460…

That’s the first step.

Once you update your drivers, you have to absolutely put PL to 360 otherwise the card would not take any mclock change.

Once you upgrade the driver version and put the power limit to 360, you would be able to apply your OC.

cclock : 1100
mclock : 2350 (test for 24h before try up)
fan : 80
Power Limit : 360

[Rscupra]

I dont have the “Force P0 State” option.

[alkilo]

the Force P0 State is useless, don’t check it, and if you don’t have it, it doesn’t matter.

[gc999999]

I get 121.1Mh with 1100, 2300, 360. Key is PL360, the real power consumption is 314W, but seems quick stable now, will keep testing for the next 24 hours.

[schneizel1208]

Here’s my settings.

I should also point out that I have an active heatsink that is thermal padded to the backplate of my 3090 evga ftw3, in an air-conditioned room at 21C. I have also ran this rig in an open room with fan ventilation without any issue. Always ran 121MH stable. Never crashed.

EDIT: Maybe try reducing the core clock? I am running it at -200 setting. I think I vividly rmb the hash rate got worse with core overclocked

[gc999999]

evga ftw3 only, try this one, fan=80; core=1100; mem=2200; PL=0
stable @121Mh, and power is around 290W.

[tisoloteam]

Hello,
What is the temp of the card ?

[tisoloteam]

Hello,
Could you share, please, what is the setting you have between -200 and 2200 on the screenshot ?
Do you think 90% is okay for the card in a long run ?
Did you try to lower the power ? To me 330W is a lot a bit…

[tisoloteam]

Hello,
When you set 360W how did you determine that the real consumption is lower ?
Using a power greed meter ?

Do you think that having 300W configured the EVGA card will actually consume less ?

[jpljpl123]

Ohhh my god thank you Soo much my brother you fix my issue i pass 2 week on it I change my psu, the riser, cable and the only solution i have its to plug my riser on sata and it’s work but not good for fire now its work with the pcie 6 pin on my riser with the 360 PL thank you so much again