What's new
Carbonite

South Africa's Top Online Tech Classifieds!
Register a free account today to become a member! (No Under 18's)
Home of C.U.D.

Power hungry 3080ti

AlexCT

Senior Member
Rating - 100%
11   0   0
Joined
Jul 6, 2015
Messages
87
Reaction score
18
Points
2,935
Hi there,

Quick brief my setup is as follows:

ASUS ROG 750W PSU
i7 10700k - turbo on
4 x 8GB 4266hz DDR4
MSI Ventus 3080ti

I have just bought the 3080ti off carb. So when I installed and got into Call of Duty Modern Warfare or as I got into a game my pc would instantly shutoff and reboot.

At this stage I was running a single 8pin from the PSU to the GPU and daisy chained it. So I thought ok maybe she's super power hungry so I ran another 8pin direct from the PSU. Low and behold no crashing - plays fine for hours.

Then since I came from a 1080ti I decided to bump up my graphics and enable ray tracing. Boom instant shutdown and reset. Turned graphics back down and can carry on playing for hours.

Then I thought I should run a stress test with Furmark and it runs fine. Then while running furmark I open cod on low settings and instant crash.

So my thinking is... either some memory on the 3080ti is bad or my PSU is bad.

PSU meats minimum recommended requirements on both the card on the psu guide. GPU never goes above 72degrees.

What can I do to diagnose... any thoughts?
 
Definitely sounds like a PSU issue

When I got my 6900xt, my (albeit old) 750w Corsair PSU started making a ticking noise under load and the fan would ramp up to full. I never got any shutdowns, but that PSU was struggling.
The 3080ti draws more power than my Radeon.. so it very well could be an overload.
 
Memory issue on the card would cause you artifacting long before a hard shutdown.

A shutdown/reset like that is power related for the most part.
Can you run HW monitor or the like to check power draw under load?
 
my 3080ti HoF draws 420w at full load

test it by having all your graphics settings on cranked as shit then change your power limit to 80% in afterburner

Do this all the way to 60%

If youre still getting crashes then idk

If it stops crashing then try an 850w psu
 
Memory issue on the card would cause you artifacting long before a hard shutdown.

A shutdown/reset like that is power related for the most part.
Can you run HW monitor or the like to check power draw under load?

Where does HW monitor show the wattage draw?
 
my 3080ti HoF draws 420w at full load

test it by having all your graphics settings on cranked as shit then change your power limit to 80% in afterburner

Do this all the way to 60%

If youre still getting crashes then idk

If it stops crashing then try an 850w psu

So the power draw options will limit power to the GPU?
 
yes, i was able to run the card on a 650w by dropping to 60% while waiting for my 850w to arrive

Alright will try this. And then do you think I should RMA the PSU? Or is it just too much power for it?

Was running 100s with the 1080ti
 
This might sound suuuuuuper off-topic, but I noticed you own a Z490-H Strix and you're running 4X8 4266C19. I would highly suggest you download HWiNFO and make sure that the board didn't auto VCCSA and VCCIO to 1.6V. It has happened and it has indeed fried IMCs.
 
Weird.. you don't get Wattage readings.

Maybe try afterburner?

rzpCB3E.png
 
You fixed the first issue by not daisy-chaining the 8Pin PSU cables. You will need 2x 8pins direct from PSU.
Also, use HWInfo64 for checking sensors. I think you will need to configure CPUID to change from % to W.
If it were memory, you would notice artifacting. Straight resetting is due to overload protection.
Verify Temps & Power usage of GPU. I would think with the 10700K & 3080Ti you would need a 1000W to be comfortable.
 
The 10700k doesn't draw that much power.. I think around 150w on average.
That GPU will draw 350w to maybe 400w at full load if overclocked or an OC version.

@AlexCT
Go here and add in all your hardware.


The GPU plus CPU you have will pull maybe 550W max, which leaves you headroom.. but all your other stuff pulls power too and you could be nearing max load on the psu.
 
You fixed the first issue by not daisy-chaining the 8Pin PSU cables. You will need 2x 8pins direct from PSU.
Also, use HWInfo64 for checking sensors. I think you will need to configure CPUID to change from % to W.
If it were memory, you would notice artifacting. Straight resetting is due to overload protection.
Verify Temps & Power usage of GPU. I would think with the 10700K & 3080Ti you would need a 1000W to be comfortable.
Tbh my 750W handled an EVGA 3080 Ti FTW with a 10850K pretty well. It even handled a 450W 6900 XT and a 1.45V 12900K in Time Spy Extreme. I honestly don't know if it could be chalked up to the PSU.
 
I am still leaning to PSU.. The straight up restarting or turning off is power related.
I don't think he's hitting the max power output.. Could be a faulty PSU that is hitting over current protection early or any other myriad of reasons.
 
Hi there,

Quick brief my setup is as follows:

ASUS ROG 750W PSU
i7 10700k - turbo on
4 x 8GB 4266hz DDR4
MSI Ventus 3080ti

I have just bought the 3080ti off carb. So when I installed and got into Call of Duty Modern Warfare or as I got into a game my pc would instantly shutoff and reboot.

At this stage I was running a single 8pin from the PSU to the GPU and daisy chained it. So I thought ok maybe she's super power hungry so I ran another 8pin direct from the PSU. Low and behold no crashing - plays fine for hours.

Then since I came from a 1080ti I decided to bump up my graphics and enable ray tracing. Boom instant shutdown and reset. Turned graphics back down and can carry on playing for hours.

Then I thought I should run a stress test with Furmark and it runs fine. Then while running furmark I open cod on low settings and instant crash.

So my thinking is... either some memory on the 3080ti is bad or my PSU is bad.

PSU meats minimum recommended requirements on both the card on the psu guide. GPU never goes above 72degrees.

What can I do to diagnose... any thoughts?

Sounds like your GPU is spiking in usage and causing overcurrent protection to kick in.

It’s not that 750watts ain’t enough, it’s that the PSU tries to protect itself too aggressively.


I’m using a 650 watt Corsair TX with my 3080 and it’s fine.

Here:

 
Last edited:
My 3080 msi trio z ran on 605watt evga G2, but ya guess you probably need more juice for the Ti!
 
I found a review and OPs PSU can happily do 841W, which is plenty.

Not sure what the issue is. Might be transient response but this is a well made and well reviewed PSU so not sure.

 
I found a review and OPs PSU can happily do 841W, which is plenty.

Not sure what the issue is. Might be transient response but this is a well made and well reviewed PSU so not sure.

841w at the wall from a PSU with 90% efficiency at full load gives you +- 750w.

Seems to me they measured power consumption of PSU not output
 
841w at the wall from a PSU with 90% efficiency at full load gives you +- 750w.

Seems to me they measured power consumption of PSU not output

Unless they just haven’t communicated their test that well and the PSU is drawing 925 watts from the wall and delivering 841 to the system.

I doubt that they would be highlighting the psu drawing it’s rated wattage as a pro in the end comments.

Pretty sure this psu can deliver 841 watts to the system.

It might be that the psu doesn’t like the sudden increase in load created by the 3080 Ti. A spike may cause the Psu to go into protection mode and shut down. The psu might prefer slow increases in power draw rather than spikes.
 

So here’s the thing, I kinda already guessed it from the initial post, but he has a 350W card. His 10700K under a gaming load doesn’t draw more than 90-100W; my 12900K doesn’t do more than 120W in game on account of the fact that it never goes above 65c, and it won’t be able to maintain under 70c at 150W or above. Other than that, I’ve owned a 10700K and a 10850K and observed their power behaviour in game. That’s to be expected because games aren’t thaaaaat CPU bound.

By my rough calculations, his system shouldn’t require more than 500W at load in COD, even if he had all his settings cranked. It might be something as simple as the 350W being draw from two PCIe cables instead of three, or perhaps one of the cables just isn’t inserted properly (I’ve had that happen as well). But yeah, if his 750W can’t do 500W that would be super weird. The only other thing I can think of is if he has like 9-12 fans or like two-three HDDs hogging power as well.

My guess would be that the PSU itself might be having some sort of issue, but then again, it could just be down to the power distribution from the PSU’s rails. The 750W might do better when three separate cables are hooked up. My 750W has only experienced a 3080 Ti and a 6900 XT were three 8-pin PCIE’s, and again, it’s been happy supplying 450W to either card (Crysis 2 Remastered with RT for the 3080 Ti FTW3, and Time Spy Extreme for the 6900 XT).

@AlexCT If you can, please check HWiNFO when you have time. You wouldn’t be the first person I’ve helped save their memory controller. It’s really worth checking for what it could prevent. It’ll only take a couple of minutes.
 
Last edited:
How many hard drives, how many rgbs etc

750w aint enough for that spec
 
How many hard drives, how many rgbs etc

750w aint enough for that spec
COD MW won't utilize a 3080 Ti to its full capabilities, the game doesn't really scale that well. But again, even if it could, we're talking 500/750W. Are PSUs not able to handle 2/3 of their rates capabilities? I think the issue is elsewhere, and I could very well be wrong. I'd just be really surprised if his 750W wasn't good for 400-500W game load. It would give me a really bad impression of the ROG branded PSUs. The 10700KF at 1.4V (I've tested it) doesn't move above 100-110W under a gaming load, and unless he BIOS modded his card, 350W is all he's getting out of it under a worst-case scenario.
 
All i can say is I went from a 700w to a 1000w on my pc with similar specs. My stability issues dissappeared
 
All i can say is I went from a 700w to a 1000w on my pc with similar specs. My stability issues dissappeared
You have a HOF though, and I assume not an i7? It could be that the extra 100W on your card and possible 20-30W on the CPU were pushing your 700W too far.
 
You have a HOF though, and I assume not an i7? It could be that the extra 100W on your card and possible 20-30W on the CPU were pushing your 700W too far.
Yep for sure. To the OP, if you can borrow a psu to test. I cant think of another reason you would have instability
 
Yep for sure. To the OP, if you can borrow a psu to test. I cant think of another reason you would have instability
I have a sneaking suspicion it has something to do with the RAM. Black screens due to lack of uncore voltage have happened to both myself and a friend, although I definitely think the single eight-pin was the reason for the initial crashes. If his IMC has degraded due to high stock voltages, it would've decreased his top-end data rate, and 4266 on Comet Lake isn't that easy to do. An easy way to check would be if OP checked disc cleanup for memory dump files.
 
I have a sneaking suspicion it has something to do with the RAM. Black screens due to lack of uncore voltage have happened to both myself and a friend, although I definitely think the single eight-pin was the reason for the initial crashes. If his IMC has degraded due to high stock voltages, it would've decreased his top-end data rate, and 4266 on Comet Lake isn't that easy to do.
Definitely need more than 1x 8pin
 

Users who are viewing this thread

Latest posts

Back
Top Bottom