XFX 7900XT hotspot issues

[farookhirani]

Hi,
does ne one with an XFX 7900XT have hotspot issues with their cards.

 

[mama]

Which specific model - the reference or custom design?

 

[AsRock]

Hi,
does ne one with an XFX 7900XT have hotspot issues with their cards.

Some of the reference XTX cards did ( manufacturing defect ) but it should of been resolved by now i would of thought.


You could all so check here

AMD RX 7000 series GPU Owners' Club

Oh I'd bet on finewine happening. I'm just not sure it's worth all this. 6700XT and 6750XT pretty much finewine'd their way past the 3070 and 3070 Ti in a bunch of games, but I never cared about that - I just wanted to get away from 8GB of VRAM. On a single monitor, with a clean Driver Only...

www.techpowerup.com

All so check the reviews that are on this site.

 

[ratirt]

Hi,
does ne one with an XFX 7900XT have hotspot issues with their cards.

Can you elaborate on the issue? What exactly is the issue with the hot spot?

 

[farookhirani]

XFX is using the reference design so the heatsink is the same as the amd cards.

 

[abysal]

This was from today's gaming session, fans were blasting full. I think its the first time it did this. This is the XFX 7900XTX MERC 310 model. Ambient temp in the room is 75F, the case is a Corsair 5000D Air with 3 x 120mm front intake @ ~1200 RPM and 1x 120 exhaust.

GPU Temp is 61C, but hotspot hitting 110C... something is wrong me thinks. Card has been in service since Feb 2023. Its running at stock settings. AMD drivers 23.7.1. Weird I thought I was on 23.7.2... I guess I'll upgrade.

 

[Space Lynx]

@abysal i would recommend ddu clean and restart. Re-install amd chipset drivers with latest from amd website, reinstall gpu drivers, make sure windows is fully up to date, reboot. Then run time spy or a demanding game amd report back your temps then without messing with fan curve at all. I'm curious to see what your temps are after all of that.

You could also flip the switch on the gpu to try the other bios. Probably won't change anything but worth a try.

 

[evernessince]

Some of the reference XTX cards did ( manufacturing defect ) but it should of been resolved by now i would of thought.


You could all so check here

AMD RX 7000 series GPU Owners' Club

Oh I'd bet on finewine happening. I'm just not sure it's worth all this. 6700XT and 6750XT pretty much finewine'd their way past the 3070 and 3070 Ti in a bunch of games, but I never cared about that - I just wanted to get away from 8GB of VRAM. On a single monitor, with a clean Driver Only...

www.techpowerup.com

All so check the reviews that are on this site.

Correct, there was a bad batch with too little liquid in the chamber. Haven't heard of another defective batch since. Hard to tell if OP is asking about that specific issue or if they are speaking more broadly, as just in general if people are having too high hotspot temps. Not enough effort was put into the opening post so people are having to waste time questing what the OP was asking.

 

[Space Lynx]

The other poster abysal has a non reference model hitting 110 Celsius, which i find very strange.

 

[LabRat 891]

While I disagree with this content creator's opinion of "don't buy AMD", he does demonstrate the (likely/inter-related) cause here:

 

[Space Lynx]

@abysal what's your gpu sag like, did you install the support bracket? I have the support bracket plus a piece of wood cut out perfectly to gove the very end of the support bracket a 0.5mm lift, to prevent any possibility of future sag.

 

[Vya Domus]

I have a reference 7900XT, with the stock cooler it would hit 100C but only overclocked with like 30C room temperature.

While I disagree with this content creator's opinion of "don't buy AMD", he does demonstrate the (likely/inter-related) cause here:

He does not know what he is talking about, either the die/IHS or the cold plates are designed to be convex like that on purpose, it's meant to improve the contact towards the core of the die where most of the heat comes from. If you take apart 99% of cards out there you'll see the same thing, the paste is spread out from within the center, if it wasn't like that you'd get gaps.

 

[abysal]

Hey guys, thanks for all the replies.

I ran time spy for 6 - 7 loops and it hit 107C on the hotspot. I only re-installed the GPU driver last night. I did not DDU. I'm on the latest chipset and latest windows 10 updates. I have a support stand, I'll check and make sure its tight and hasn't shifted and that the card is level. I could see how the sag of this heavy card could mess up the contact with the GPU die. I also found my baseline test when I first got the card - although I don't remember what software I used to test it, but its dated 3-2-2023 and the hotspot temp was 92C, with similair max GPU temp as currently.

I think I'll try a re-paste this weekend... these temps definitely look like they got worse since when the card was first installed. My clocks are considerably lower now too because of this temp issue... Hopefully its Sag or paste related. Otherwise... reach out to XFX for RMA?

Original Temp Testing - March 2023

Current Temp Testing - August 2023 - 23.7.2 Drivers

Build w/ Sag Support & Full System Specs:

CPU: AMD Ryzen 9 7900X 12-Core Processor - Phanteks PH-TC14PE Cooler -> Testing Ryzen 5 7600x 6-Core Processor @ 5.5 GHz (CPU Game Mode) + 1.200v SoC
RAM: G.Skill Tident Z5 Neo DDR5-6000 CL32-38-38-96 1.40v 64GB (2x32GB - F5-6000J3238G32GX2-TZ5NR). Also works: Corsair CMK32GX5M2B5600C36 32GB (2x16GB ) @ XMP + 6000MT/s Set
GPU: XFX 7900XTX
Main Board: MSI MPG X670E CARBON WIFI (MS-7D70) - BIOS v186 - Code 00 on reboot or sleep resume
Storage: Crucial P5 Plus 2TB - Crucial P3 4TB - addlink s70 2TB - SP A80 1TB - Seagate Exos 10TB Raid1 - Samsung 870 EVO 4TB SSD
Network: TRENDnet TN9310 SFP+ 10GbE
Audio: On board ALC4080 Realtek USB Audio
PSU: Corsair HX1200i
Case: Corsair 5000d Aitflow (3 intake 120s, 1 exhaust)
Monitor: Dell 3223D @ 144Hz
Input: Razer Basilisk X Hyperspeed, Techware Phantom L TKL
OS: Win 10 - Mint 21

 

[Space Lynx]

You honestly have me a bit worried about my own 7900 xt, I already saw about a 7-8 celsius temp increase over my baseline testing. I hit 90 celsius hotpsot when I play shadow of the tomb raider, but when I first got the card 3 weeks ago my first benchmark runs in Shadow were only hitting 83 hotspot, everything else stayed the same, no variable changes.

Honestly not sure what to make of it, cause 90 technically still isn't bad... but if this keeps getting worse...

have you tried msi afterburner and rivatuner to see your temps in game? maybe your temp in-game actually isn't that bad, but its just spiking to 107 celsius in one scene or something.

problem with repaste is xfx has the warranty sticker over one of the main screws, and while technically they still have to honor warranty if that is removed, its not something i want to risk personally, cause they could always make up some other reason to not honor if you piss them off by removing the warranty sticker for a repaste. just depends who is working that day and their emotional state you know?

you got me worried now though, cause guru3d review of my exact merc310 7900 xt card says hotspot was only 74 celsius for his review, and he tested games likes shadow of tomb raider too...

fuck, I wonder if something is wrong with our vapor chambers

 

[abysal]

You honestly have me a bit worried about my own 7900 xt, I already saw about a 7-8 celsius temp increase over my baseline testing. I hit 90 celsius hotpsot when I play shadow of the tomb raider, but when I first got the card 3 weeks ago my first benchmark runs in Shadow were only hitting 83 hotspot, everything else stayed the same, no variable changes.

Honestly not sure what to make of it, cause 90 technically still isn't bad... but if this keeps getting worse...

have you tried msi afterburner and rivatuner to see your temps in game? maybe your temp in-game actually isn't that bad, but its just spiking to 107 celsius in one scene or something.

problem with repaste is xfx has the warranty sticker over one of the main screws, and while technically they still have to honor warranty if that is removed, its not something i want to risk personally, cause they could always make up some other reason to not honor if you piss them off by removing the warranty sticker for a repaste. just depends who is working that day and their emotional state you know?

you got me worried now though, cause guru3d review of my exact merc310 7900 xt card says hotspot was only 74 celsius for his review, and he tested games likes shadow of tomb raider too...

fuck, I wonder if something is wrong with our vapor chambers

I've dealt with XFX before and while not as good as EVGA (gang-bang'in thugs who aint with us no more), they do specifically state that breaking the warranty sticker in US and CANADA will not void your warranty if you decide to re-paste and clean fans, provided you return it to how it was before and don't cause any damage.

Its interesting that we both see much high temps than in the reviews. I was also surprised when my hotspot was 92C originally, but I figured the the clocks are high and its within spec. The reviews had the hot spot at 70 - 85C, I'd have to look it up so I'm not pulling the numbers out of my ass.

I'm going to follow your suggestion and install rivatuner and monitor temps that way, and also run a few rounds of MW2.

 

[Space Lynx]

I've dealt with XFX before and while not as good as EVGA (gang-bang'in thugs who aint with us no more), they do specifically state that breaking the warranty sticker in US and CANADA will not void your warranty if you decide to re-paste and clean fans, provided you return it to how it was before and don't cause any damage.

Its interesting that we both see much high temps than in the reviews. I was also surprised when my hotspot was 92C originally, but I figured the the clocks are high and its within spec. The reviews had the hot spot at 70 - 85C, I'd have to look it up so I'm not pulling the numbers out of my ass.

I'm going to follow your suggestion and install rivatuner and monitor temps that way, and also run a few rounds of MW2.

also are you letting fps go full throttle? or are you using fps cap/vsync something like that? in shadow of tomb raider benchmark i get 3 celsius less hotspot with fps cap on. not much, but seeing 86 or so vs 90 is nice. but again the guru3d review showed my card (xt not xtx) doing 74 hotspot at stock... so eh still concerning.

edit: @abysal try flipping the switch for the other BIOS on the XTX, i just did this and it dropped my temps 10 celsius.

 

[abysal]

I've set my fps cap to 140 max and adaptive sync on. And in menu's I limit frames to around 60 if I can. So I logged with AMD Adrenalin, its really decent software. And also watched my overlay over the course of 6 to 7 matches in MW2 across several different maps. I seen it spike in the overlay to 96C, with maybe 350watts board power. Most of the time it was in the mid to high 80s. And I never heard the fans full blast. I think you maybe right that its just briefly spiking to those high temps. I don't know if that's normal...

Here's the CSV log file, which honestly doesn't look bad at all. See if you can log yours and gain a better view of temps.
https://docs.google.com/spreadsheets/d/1hdVuoPMmQ0_N_uYiaZLm-igSMqgMJh8frC9kbemXWog/edit?usp=sharing


I'll do some more logging in Time Spy over the weekend. But I think I'll hit YouTube university and see how these cards come apart and I'll apply some new paste. I'll see if I can set the BIOS switch to the other position, and see what the temps look like, is that a high fan curve or no auto off fan? Can you do that on the fly or do you have to power down to switch it and for it to register?

 

[Space Lynx]

I've set my fps cap to 140 max and adaptive sync on. And in menu's I limit frames to around 60 if I can. So I logged with AMD Adrenalin, its really decent software. And also watched my overlay over the course of 6 to 7 matches in MW2 across several different maps. I seen it spike in the overlay to 96C, with maybe 350watts board power. Most of the time it was in the mid to high 80s. And I never heard the fans full blast. I think you maybe right that its just briefly spiking to those high temps. I don't know if that's normal...

Here's the CSV log file, which honestly doesn't look bad at all. See if you can log yours and gain a better view of temps.
https://docs.google.com/spreadsheets/d/1hdVuoPMmQ0_N_uYiaZLm-igSMqgMJh8frC9kbemXWog/edit?usp=sharing


I'll do some more logging in Time Spy over the weekend. But I think I'll hit YouTube university and see how these cards come apart and I'll apply some new paste. I'll see if I can set the BIOS switch to the other position, and see what the temps look like, is that a high fan curve or no auto off fan? Can you do that on the fly or do you have to power down to switch it and for it to register?

View attachment 309624


View attachment 309623

i turned off the computer before I changed the switch on the gpu bios. but it lowered my temps by a lot surprisingly. my performance also went down, I guess I was in the OC bios and didn't realize it.