Zotac 3090 hot spot temp 105c+ causing throttling

Gimmles

Bronze Level Poster
Hi there,

As the title says, my GPU is seeing some seriously high temps reported in HWINFO when gaming.
105c and occasionally creeping above on the hot spot temps, and I can clearly see this is causing thermal throttling and reduced performance overall. (My PC is dusted regularly)

I know this is likely a hardware fault, I am wondering as I am now outside of the parts warranty period I had for my machine as to what my steps for a replacement are?

Cheers folks.

EDIT:
CaseNZXT H511 MID-TOWER GAMING CASE (BLACK)
down_right_arrow.gif
Change to: COOLERMASTER MASTERBOX TD500 MESH ARGB GAMING CASE
Processor (CPU)AMD Ryzen 9 5950X 16 Core CPU (3.4GHz-4.9GHz/72MB CACHE/AM4)
MotherboardASUS® TUF GAMING B550-PLUS (DDR4, USB 3.2, 6Gb/s) - ARGB Ready!
down_right_arrow.gif
Change to: ASUS® TUF X570-PLUS GAMING (USB 3.2 Gen 2, PCIe 4.0, CrossFireX) - ARGB Ready!
Memory (RAM)64GB Corsair VENGEANCE DDR4 3200MHz (4 x 16GB)
down_right_arrow.gif
Change to: 32GB Corsair VENGEANCE DDR4 3600MHz (4 x 8GB)
Graphics Card24GB NVIDIA GEFORCE RTX 3090 - HDMI, DP
1st M.2 SSD Drive512GB PCS PCIe M.2 SSD (2000 MB/R, 1100 MB/W)
down_right_arrow.gif
Change to: 1TB PCS PCIe M.2 SSD (2200 MB/R, 1500 MB/W)
1st Storage Drive2TB SEAGATE BARRACUDA SATA-III 3.5" HDD, 6GB/s, 7200RPM, 256MB CACHE
DVD/BLU-RAY DriveNOT REQUIRED
Power SupplyCORSAIR 750W TXm SERIES™ SEMI-MODULAR 80 PLUS® GOLD, ULTRA QUIET
down_right_arrow.gif
Change to: CORSAIR 850W RMx SERIES™ MODULAR 80 PLUS® GOLD, ULTRA QUIET
Power Cable1 x 1 Metre UK Power Cable (Kettle Lead)
Processor CoolingNoctua NH-U14S Ultra Quiet Performance CPU Cooler
down_right_arrow.gif
Change to: Corsair H115i RGB PLATINUM Hydro Series High Performance CPU Cooler
Thermal PasteARCTIC MX-4 EXTREME THERMAL CONDUCTIVITY COMPOUND
Extra Case Fans3x CoolerMaster SickleFlow 120 ARGB + Controller Kit
down_right_arrow.gif
Change to: 1x 120mm Black Case Fan (configured to extract from rear/roof)
Sound CardONBOARD 6 CHANNEL (5.1) HIGH DEF AUDIO (AS STANDARD)
Network Card10/100/1000 GIGABIT LAN PORT (Wi-Fi NOT INCLUDED)
Wireless Network CardWIRELESS INTEL® Wi-Fi 6 AX200 2,400Mbps/5GHz, 300Mbps/2.4GHz PCI-E CARD + BT 5.0
USB/Thunderbolt OptionsMIN. 2 x USB 3.0 & 2 x USB 2.0 PORTS @ BACK PANEL + MIN. 2 FRONT PORTS
Operating SystemNO OPERATING SYSTEM REQUIRED
Operating System LanguageUnited Kingdom - English Language
Windows Recovery MediaNO RECOVERY MEDIA REQUIRED
Office SoftwareFREE 30 Day Trial of Microsoft 365® (Operating System Required)
Anti-VirusNO ANTI-VIRUS SOFTWARE
BrowserGoogle Chrome™
Warranty3 Year Silver Warranty (1 Year Collect & Return, 1 Year Parts, 3 Year Labour)
 
Last edited:

SpyderTracks

We love you Ukraine
Hi there,

As the title says, my GPU is seeing some seriously high temps reported in HWINFO when gaming.
105c and occasionally creeping above on the hot spot temps, and I can clearly see this is causing thermal throttling and reduced performance overall. (My PC is dusted regularly)

I know this is likely a hardware fault, I am wondering as I am now outside of the parts warranty period I had for my machine as to what my steps for a replacement are?

Cheers folks.
Could you post your full specs from the order page please?
 

B4zookaw

VALUED CONTRIBUTOR
Have you changed your GPU fan profile at all? When GPU is running at load and reaches hot spot, what rpm or % are the GPU fans running at? And have you over clocked the card at all?
 

SpyderTracks

We love you Ukraine
Also, can you take a screenshot of HWMonitor or something showing the throttling flag?

Hotspot temp will always be high, that's expected, there's nothing unusual about that.

What's making you think it's throttling? Any screenshots would help.
 

Gimmles

Bronze Level Poster
Have you changed your GPU fan profile at all? When GPU is running at load and reaches hot spot, what rpm or % are the GPU fans running at? And have you over clocked the card at all?
I have changed the fan profile to kick in earlier than the stock that the card came with, and I have not overclocked the card at all.

when was the last time you gave it a dust out and cleaned all the filters
I dust out the PC regularly, it is fairly clean in the system at all times!

Also, can you take a screenshot of HWMonitor or something showing the throttling flag?

Hotspot temp will always be high, that's expected, there's nothing unusual about that.

What's making you think it's throttling? Any screenshots would help.
Hotspot temps shouldn't be 105c though surely, this particular temperature is clearly causing the GPU to not even be able to reach it's 83c temperature limit set by the GPU and shown in afterburner- I have posted a screenshot of HWINFO and MSI Afterburner- the GPU will never go above 70-73c at most, but the core clock slowly starts creeping down while gaming over longer periods.

Whilst gaming at longer periods, the Hot Spot temp will start hitting 107c-108c and the tachometer for the fans will show a spike to 3000~ and the computer will get pretty loud for about 10s or so, this will repeat over and over. (I have a second picture showing this)

I can't show you the degraded game performance, but it's a noticable difference from the FPS/clock speeds I used to get.

Edited to add that MSI afterburner also seems to think the card is hitting it's temp limit- which to me is another clear indicator that something is very wrong (3rd image)
 

Attachments

  • e0783a3408b79e4276b1868cda274a22.png
    e0783a3408b79e4276b1868cda274a22.png
    950.1 KB · Views: 282
  • img2.png
    img2.png
    935.8 KB · Views: 292
  • temp limit.png
    temp limit.png
    951.8 KB · Views: 270

SpyderTracks

We love you Ukraine
I can't show you the degraded game performance, but it's a noticable difference from the FPS/clock speeds I used to get.
We need to know the clock speed difference and FPS difference, but there's nothing untoward about those hotspot temps, they will always be very high.

It would be worth running a furmark stress test for say 10 minutes and registering temps with something like HWMonitor, or you can capture gpu sensor graphs with something like GPU-Z while it's running.

But I don't see anything untoward from those readings personally, where are you seeing it's hitting it's limit? I don't see that anywhere, in fact is over 5 degrees below it's limit which is really healthy.

The hotspot is not a relevant reading to be concerned about, if you're thinking that should be below the TGP of 85ish, that's not the case, hotspot readings will always be extremely high.

The fans ramping is a known bug on those GPUs, you can disable the feature.
 
Last edited:

Gimmles

Bronze Level Poster
We need to know the clock speed difference and FPS difference, but there's nothing untoward about those hotspot temps, they will always be very high.

It would be worth running a furmark stress test for say 10 minutes and registering temps with something like HWMonitor, or you can capture gpu sensor graphs with something like GPU-Z while it's running.

But I don't see anything untoward from those readings personally, where are you seeing it's hitting it's limit? I don't see that anywhere, in fact is over 5 degrees below it's limit which is really healthy.

The hotspot is not a relevant reading to be concerned about, if you're thinking that should be below the TGP of 85ish, that's not the case, hotspot readings will always be extremely high.

The fans ramping is a known bug on those GPUs, you can disable the feature.
There's no way 105c+ is acceptable for hot spot temps, everywhere else I am looking tells me that it is way too hot, you can see at the bottom of the 3rd screenshot in afterburner that temp limit is at 1 and not 0.

Moreover, the default boost clock is 1695. If the card is only at 70c and not throttling, why is the boost clock dropping over time (you can see its not staying at that boost even in those screenshots) when that clearly implies the card is reducing its performance.

I will do some stress testing on the card when I am home later. Would also like information on how to solve the fan ramp up bug, if possible. Thanks
 

SpyderTracks

We love you Ukraine
everywhere else I am looking tells me that it is way too hot, you can see at the bottom of the 3rd screenshot in afterburner that temp limit is at 1 and not 0.
Have you got links to those sources?

If you could run a frmark test which is a known load and record the GPU and CPU (both package and core) with GPU-z (output to file)

Run that for 10 minutes and then we've got a known baseline to relate to.

I would disable afterburner and HWMonitor during this time as it may well conflict for the sensors.
 

Gimmles

Bronze Level Poster
Have you got links to those sources?

If you could run a frmark test which is a known load and record the GPU and CPU (both package and core) with GPU-z (output to file)

Run that for 10 minutes and then we've got a known baseline to relate to.

I would disable afterburner and HWMonitor during this time as it may well conflict for the sensors.
Hi there,

Not 100% sure how to have effectively benched with this program, but I've attached the sensor log from a 10 minute run on the GPU and the GPU Shark app after 10 minutes of running.

EDIT: I uploaded the sensor log as a txt but it works better if you just change the file extension to CSV- makes the data more readable
 

Attachments

  • GPU-Z Sensor Log.txt
    462.6 KB · Views: 126
  • gpu shark.png
    gpu shark.png
    48.6 KB · Views: 184
Last edited:

SpyderTracks

We love you Ukraine
Not 100% sure how to have effectively benched with this program
Ok, yeah, the Hotspot is immediately ramping straight up as soon as the load it applied and hitting a ceiling suggesting throttling.

Couple of options here:

1/. If you feel confident, it's likely just a repaste job required on the GPU die, possibly some extra well placed thermal pads also, but there's no way of knowing that until you remove the shroud and see what's applied. But the Memory temps look reasonable, so I don't believe thermal pads would be required, it's just that hotspot that appears to be out of hand.

2/. Manufacturers warranty is still in place, so you're fully covered by Zotac.

Your card is the ZOTAC GAMING GeForce RTX 3090 Trinity OC:


The basic warranty on GPU's from Zotac is 3 years:

Graphics Cards (Only RTX 30 Ampere Series)

· Standard Warranty: 3 years

· Extended Warranty: +2 years

· Total Warranty: 5 years total


Unless you're happy removing the shroud (which can be quite daunting if there are small cable connects that need disconnecting internally before removing the shroud), I would recommend RMA'ing the card, just covers yourself

Just to put your mind at rest, Zotac have a good RMA process generally, it may take a few weeks, but they'll likely just replace the card. There's a thread of people's experiences here:



 
Last edited:

SpyderTracks

We love you Ukraine
Forgot to post the warranty page for Zotac

They have an online chat which may be the easiest way to get an RMA sorted:


Or there's a web form.

You'll need the GPU serial number which is on the backplate of the GPU:

1649025019025.png


Best to register it first if you haven't already:

 
Last edited:
Top