I have a 3090 that thermal throttls down to 1800 mhz at 70c. i am using GPUz sensors prefcap reason to test. i have tried using asus gpu tweak to change thermal limit ( it should be 81 anyway ) this makes no difference i have tried using MSI afterburner this also made no difference. the card WILL NOT go above 70c. it is driving me insane.
#3090 Throttle at 70c
1 messages · Page 1 of 1 (latest)
anyone?
GPU below 80-90 is fine
after 80 it begins thermal throttling
also why downclock ram? you'll lose gpu performance
Likely stale presets that auto load
uninstall both Asus GPU Tweak and MSI Afterburner, deleting all settings if prompted
Restart the PC
Then, download Display Driver Uninstaller and reinstall graphics drivers
Check if you don't throttle at 70C, but rather at 83C+ or never
it wont go above 70 it is throttling at 70 and the ram is running at 6800 because thats the most my system can run haha
sounds like a plan
Did it improve?
Same did it all
no change, did a reinstall of windows. everything clean fresh install of the GPU drivers. didnt install anything with the ability to mod said GPU curve or voltages or temps nothing. still the same
still throttling down at 69 and wont go above it
Hmm...
even if i have aircon running at my PC right so it wont thermal look at this proformance
thats a 3090 being 100% of a bottleneck
this is a benchmark
other 3090s on this benchmark for warzone are showing around 250-280 fps
its driving me insane
i have hwinfo
While doing the thing that causes the GPU to throttle, look at the contents of it
Specifically, the entire GPU section
therm
screenshot of the entire section
okay will do
You're in the wrong mode. Click sensors at the top left corner of the window
im back got called away for work
Did you figure out what's happening at least?
nope
will do the print screen today
its got worse now getting fragments when its thermaling
@subtle heath thanks
100C hotspot vs 65C nominal? Yeah, cooling's a bit sussy here
Is this thing still in warranty? If not, you're going to need to do something to the cooler. Likely a repaste since the GPU die isn't making the best contact with the cooler.
Looks like the mystery was solved at least. The GPU average temp is fine, but the hotspot is way too high
there is the issue
the everage temp is not fine
it WONT go above 69
and thermal throttles at 69 average temp
instead of the 81 its meant too
ya its in warrenty hence why i have not re thermal pasted and pad on there but instead of waiting weeks for an RMA i wanted to make sure what it was
i knew it was over heating, the issue is the thermal trottling at 69 and not being able to go above it @subtle heath
I think you missed the part where the hotspot is abnormally high
Bets are the card won't respect the normal thermal limits if other parts were cooking
righttt
@subtle heath Scan have said these are normal temps... will see how my RMA goes...
@flint peak have you checked the GPU-Z sensors tab for the throttle reason?
It's called Perfcap
I assume it will show a temp reason. And you can tell scan to stop being daft.
A delta that big between the core average and the hotspot points to something being wrong. And with a hotspot that high it's setting your fans to 100%
It never does that unless it's overheating.
108c is hot enough to damage the core tbh
yep therm
schedualed to send back monday i did all these tests few weeks ago told them the readings and the term prefcap and he said 108 hot spot is normal you do know its hotter then the rest " the 12 years system engineer in me wanted to slap him through the phone "
i only asked in here JUST in case it was something dumb af i was overlooking. you guys have been great
Is there anything i can do if they come back and say its not faulty?
If you are sending it back to scan, they should do an initial check and then either replace it or send it back to the manufacturer to RMA it
I don't see how you would conclude that card doesn't have an issue
Because with that size delta, I think even if they put it on an open test bench. If they run a heavy workload it's going to go to 100% fan speed, which isn't normal.
well fingers crossed as over the phone they said temps were normal we will see
thanks again
having an argument atm with them telling me the temp does not get above 59 i asked what tests they are doing and how its being done - "open bench test running spiderman remastered "
sigh this is hard work
Ask them to run the GPU tests on OCCT. Although if they haven't paid for it they may not be allowed to.
And that it's the hotspot temp not the core temp.
And if they are still being useless you can just request they RMA it back to the manufacturer.
I have requested more information on what is being used to test the card as i dont understand how using spiderman is an actual test.
as i will dispute this if they tell me its working fine
after 3 days they finally ran ooct and immidiatly found faulty after spending 2 days running spiderman....
They will have to yeah. Depending on how long ago it was bought though they might have to send it off to the manufacturer first.
4090 replacement
replaceing the 3090 with a 4090
Damn not bad lol
noice
not bad at all....
cant really ask for more...
Honestly surprised they gave a 4090
yep the same editon as my 3090 asus OC
its running at 45C underload