RTX 3090 starts at 119mh/s then drops to 98-104mh/s temps at 48-52c

I just installed hive os, I tuned up the overclock settings to some settings that where posted on the forum:Core -100 Mem 1800 Pl 295, but I can’t get over 104mh/s, it starts off at 119mh/s then drops to 98-104mh/s, and I have pretty good temps:48-52, does anybody know what the issue could be?

Thanks

I have similar problem, i ran my 3090 on 122mh for 2-3 weeks and then it started to drop to 100-115


i didn’t changed nothing and have serious drops, thought it was software problem(it started after i updated OS) so i rolled back, tried fan control but every change i make just low’s the fan speed so left it on 100%

**i have a same tissue **
**i have 2 different edition of GeForce 3090 cards **
**MSI RTX 3090 and EVGA RTX 3090 **
the MSI has stable hash rate at temperature 53 to 58 C
but the evga un stable even the same temperature , i tried more OC settings but no gain .
and also, there is a discrete hash rate on time line despite of graph of that pool i mining in .
i don’t know is that normal?! and what should i do ?

are you still having this issue? Curious about your OC and whether you’ve tried a few common 3090 OC settings that have been posted on a few topics here in Hiveos forum.

Like a few that have posted, I run the following OC on four eVga 3090s with zero issues for 120.0 MH:

1 Like

I’m running -300 2250 PL 290 at 121mh
Msi rtx 3090 suprim x v stable as long as I keep the room cool with a good air flow

Please help

is there anyone solved this problem?

Your probably thermal throttling. Your memory is way to high. The absolute core setting you have for the 3090 is a setting you should have for all the 3070s, not the 3090
I dont think there is an absolute core setting for the 3090s.
Change all ur cores on the 3070s to 1100, memory to 2575 and power limit to 125. Youll save about 10-15 watts per gpu and your temps show get cooler.

The 3090 -200 core and 2000 memory. Power limit to 320.

Your power limit is to low. These 3090s are power Hungry. Go to 320 PL

Thermal throttling 100% change the pads and you will be fine

Id try the power limit first. Ive never seen anyone get 120MH out of a 3090 without being over 300watts
Sometime absolute core doesnt work in every case.

48-52C is GPU core temp. You did not see GDDR6X temperature because nvidia has not released the sensor API for linux-based OS. Which is why you will only see one temperature on NVIDIA, while you will see two on AMD cards.

GDDR6X can hit 100C easy. What’s more on the EVGA card, there are GDDR6X memories on the bottom layer that only has the metal backplane to sink all the heat. As to why this can pass any form of safety and fire hazard regulation beats me.

I have a EVGA RTX 3090 FTW3 mining with a custom made copper blocked that was CNC’d to the dimension of the backplane…mining ETH at 115MH @310W, 1400 mem, -300 core. This is with a 240Vac fan blowing air laterally along the copper block… i have not dared to push the power beyond 330W or 2000 mem

Now u did :slight_smile:

Thats a high memory setting. I can push any gpu to its max but that doesnt mean its safe long term.

Hi
I tried your setting but not work with my Gigabyte Aorus 3090 :frowning: … any ideas?

I’m very new to mining, Am I killing my GPU’s? I understand the Gigabyte has cooling issues which impact temperatures and hashrate.

1 Like

You will kill your 3090s if you dont actively cool your backplate.

I bought a heatsink with active fans and thermal pad it to the backplate of my evga 3090 ftw3 and asus strix 3090

God knows what my memory junction temps are unless I boot to windows and pull out hwinfo64 to see the readings…

But it hasn’t throttled my hashrates so I assume that the junction temp is within 115C (which is “safe” according to Micron and Nvidia)

interestingly this seemed to have a positive effect on my worst performing 3090, thank you!

i think putting in some good pads will help more tho!
this one is even worse than my 3080 that have proper pads !

Hi I have three 3090 ewga ultra ftw3 too. But have some problems with hash rate stability.
Which settings do you use now?
Could you advice the type of radiators for the card backplate ? Can you send photo or web link?

answer: maybe, yes.
UNDERclosk your core to -300
and try memory of 2000. that should get you 120MH and it should be stable. give it an hour to stabilise with the fans cranked up.
then try 2400 if you want to or maybe 2500. i wouldn’t go over 2400 though as that will only eke out an extra MH or so and introduce instability, especially over summer…
hive os lets you look at popular settings. they are popular for a reason!! but they don’t indicate how to deal with errant cards