AMD RX 7000 series GPU Owners' Club

[Zach_01]

I have previously posted my temperate observations on my Powercolor 7900XTX Liquid Devil here. The GCD hotspot when I got it was hitting 110C. Since that time, and after advice from this thread, I have repasted twice. First was a dose of Kryonaught Extreme. No noticeable improvement. The second was PTM7950. The hotspot is still not great, but I have noticed higher water temperature which suggests to me that heat is transferring better out of the card with the second repaste.

This is my conclusion based on a sample size of one: This Powercolor model of the 7900XTX is designed to run this way. I haven't noticed any throttling and indeed the card performs well but I can't see how the hotspot temperature can be moderated while the card allows in 550W.

Before anyone asks, I have not enabled the "unleashed" bios. It's not overclocked nor is it undervolted. Perhaps I should undervolt it but I think people who are thinking about purchasing this card should be aware this.

What is the game that you get those temperatures? ...and drivers? I have 24.2.1 currently
This kind of GPU/Hotspot delta seems to me like cooler/block contact issue(s)

Can you run a common benchmark that would give you similar thermal behavior?
I'd like to try it on my nitro+
Something like Superposition maybe...

Heads up
Dont go please above 400W for TBP as I cant go beyond because of my "small" PSU, even tho the nitro+ can go up to ~465W on primary/fast BIOS

If you decide to do it try and make visible all the GPU values

This is a bit awkward view because I stitched 2 columns together
Just an example

Speaking of which, I installed the new 24.2.1 driver over the weekend, and now my idle power is back in the 30 W range (it was 10 W before).

Kind of same here.
I am seeing 10W as minimum but only for like 1-2% of idling time. Practically its around 30W (around ~15W before with a short minimum of <5W)
Didnt roll back to 24.1.1 though.

I'm thinking this maybe connected to windows updates of the past couple of weeks.
No drivers GPU installation happened by win but some other "settings" might doing this. And Im not sure if my assumption is valid for both Win10/11. I've upgraded to 11 a couple of months ago.

 

[GerKNG]

Card is dead...
crashed a LOT yesterday (nothing overclocked) and then it was no longer recognized by the drivers. while DDU the whole screen was full of artifacts and the PC shut off and refuses to power on with the card in the system.

 

[Space Lynx]

Card is dead...
crashed a LOT yesterday (nothing overclocked) and then it was no longer recognized by the drivers. while DDU the whole screen was full of artifacts and the PC shut off and refuses to power on with the card in the system.
View attachment 338115

you have more bad luck than anyone else I know when it comes to hardware. I seriously wonder if there is something wrong with the way electricity is wired in your home, or maybe there is an excess of static electricity building up in your house, etc. no idea, but between all the hardware that has died on you, including the qlc storage drives... something seems off to me.

 

[AusWolf]

Kind of same here.
I am seeing 10W as minimum but only for like 1-2% of idling time. Practically its around 30W (around ~15W before with a short minimum of <5W)
Didnt roll back to 24.1.1 though.

I'm thinking this maybe connected to windows updates of the past couple of weeks.
No drivers GPU installation happened by win but some other "settings" might doing this. And Im not sure if my assumption is valid for both Win10/11. I've upgraded to 11 a couple of months ago.

Interestingly, rolling back to 24.1.1 didn't help. There might be some remnants of the other version in the system, or I don't know.

Now, another funny thing came up: while reading/commenting here on TPU in Chrome, the card eases back into the 12-13 W range with spikes up to 25-30 W, but when I close Chrome, the VRAM clock jumps to 1253 MHz, and power to 53 W... with nothing open at all! I'm getting closer to selling this card, putting my 6750 XT, or 6500 XT or 1660 Ti back in the PC (I don't have a lot of time to game anyway) and just wait for RDNA 4.

I mean, it's a great card, it plays my games on Ultra and everything, but this idle power issue is getting on my nerves.

 

[GerKNG]

you have more bad luck than anyone else I know when it comes to hardware. I seriously wonder if there is something wrong with the way electricity is wired in your home, or maybe there is an excess of static electricity building up in your house, etc. no idea, but between all the hardware that has died on you, including the qlc storage drives... something seems off to me.

Our Outlets are ~60 year old wiring with new and tested Schuko Outlets from the late 2010s in two seperate circuits and a bunch of them are in smart home voltage/current measurement devices. (they all work perfectly fine and were tested after the installation.)

i guess i just have bad luck and too much Hardware.
i even have a list of my stuff with what i did or what happened to it

Out of 40 SSDs one TLC drive was dead after a couple hours and the handful of QLC drives degraded and broke within a couple weeks to almost a year (two of them inside two different laptops). meanhwile all other drives are in the same M.2 slots on the same motherboards, with the same PSU in the same Outlet are running fine. (some are over 15k Hours now)

Out of 56 GPUs four are dead. (1060, 5700XT, 6900XT and now the 7900XTX)

Out of 48 CPUs five are dead / faulty (1x Zen 2, 3x Zen 3 and 1x Zen 4) i don't count the 7800X3D that fried after the weird extreme overvoltage when using these RAM timing presets in ASRocks bios.

Out of 33 Motherboards just one was completely toast (B450 gaming plus) and one B550 Strix-F bricked itself after a bios update.

but this idle power issue is getting on my nerves.

two 1440p 280Hz Displays on a 7900XTX at the desktop. (this would cost me 260€ per year while it does literally nothing all day long)

 

[mama]

What is the game that you get those temperatures? ...and drivers? I have 24.2.1 currently
This kind of GPU/Hotspot delta seems to me like cooler/block contact issue(s)

Can you run a common benchmark that would give you similar thermal behavior?
I'd like to try it on my nitro+
Something like Superposition maybe...

Heads up
Dont go please above 400W for TBP as I cant go beyond because of my "small" PSU, even tho the nitro+ can go up to ~465W on primary/fast BIOS

If you decide to do it try and make visible all the GPU values

This is a bit awkward view because I stitched 2 columns together
Just an example
View attachment 338107


Kind of same here.
I am seeing 10W as minimum but only for like 1-2% of idling time. Practically its around 30W (around ~15W before with a short minimum of <5W)
Didnt roll back to 24.1.1 though.

I'm thinking this maybe connected to windows updates of the past couple of weeks.
No drivers GPU installation happened by win but some other "settings" might doing this. And Im not sure if my assumption is valid for both Win10/11. I've upgraded to 11 a couple of months ago.

The most recent idle power usage change as I understand it is as a result of the newest AMD driver. Ancient Gameplay guy (Fabio Pisco I think his name is) did a YouTube video on it.

I am not saying all 7900XTX models have the same temperate observations, I am saying this particular Powercolor 7900XTX Liquid Devil model is intended to work this way. It works and works well but with a very high delta to the GCD hotspot. The behaviour is consistent since I got the card in September last year in a variety of games: X4 Foundations, Monster Hunter World, Remnant 2, Outer Worlds, Starfield, Cyberpunk, Baldur's Gate, Days Gone etc. Depending on the game, the hit on the CPU is different but the delta on the GPU temperature remains throughout.

As I said in my earlier post, a card that takes in 550W is going to get hot. This is the default bios and default behaviour for the card. I note your card pulls in 100W less. I'll run Superposition at 4K optimised for you anyway (see the attached screenshot). The ambient temperature is 22.5C and 46% humidity. The score was 22,985 btw. This is not indicative of gaming temperatures and the delta is greater when gaming for extended periods (more than 1 hour).

I communicated with my retailer about an RMA prior to commencing the two repastes and was told that the RMA was unlikely to succeed as the card was in spec (and I it is to be fair). I even engaged directly with Powercolor to see whether this is indeed the expected behaviour, but they just don't/can't give a straight answer. This is what I get back after several tries:

Dear User,
Please ensure there is enough water flow through the RX7900XTX 24G-W/OC, and the water temperature is kept as low as possible.
If you are still worried about the abnormal temperature of the graphics card, we recommend that you return to the original place of purchase to verify.
Thanks for your inquiry from PowerColor website.

 

[Zach_01]

Interestingly, rolling back to 24.1.1 didn't help. There might be some remnants of the other version in the system, or I don't know.

Now, another funny thing came up: while reading/commenting here on TPU in Chrome, the card eases back into the 12-13 W range with spikes up to 25-30 W, but when I close Chrome, the VRAM clock jumps to 1253 MHz, and power to 53 W... with nothing open at all! I'm getting closer to selling this card, putting my 6750 XT, or 6500 XT or 1660 Ti back in the PC (I don't have a lot of time to game anyway) and just wait for RDNA 4.

I mean, it's a great card, it plays my games on Ultra and everything, but this idle power issue is getting on my nerves.

I wouldn’t sell the card “just” for that. Personally I don’t see it as much of an issue. Temperatures are low still even with 30W idle. Other than this, the nitro+ is great card and I’m happy with it. And if this idle power goes down again in future, then it’s +1.

 

[Space Lynx]

Our Outlets are ~60 year old wiring with new and tested Schuko Outlets from the late 2010s in two seperate circuits and a bunch of them are in smart home voltage/current measurement devices. (they all work perfectly fine and were tested after the installation.)

i guess i just have bad luck and too much Hardware.
i even have a list of my stuff with what i did or what happened to it

Out of 40 SSDs one TLC drive was dead after a couple hours and the handful of QLC drives degraded and broke within a couple weeks to almost a year (two of them inside two different laptops). meanhwile all other drives are in the same M.2 slots on the same motherboards, with the same PSU in the same Outlet are running fine. (some are over 15k Hours now)

Out of 56 GPUs four are dead. (1060, 5700XT, 6900XT and now the 7900XTX)

Out of 48 CPUs five are dead / faulty (1x Zen 2, 3x Zen 3 and 1x Zen 4) i don't count the 7800X3D that fried after the weird extreme overvoltage when using these RAM timing presets in ASRocks bios.

Out of 33 Motherboards just one was completely toast (B450 gaming plus) and one B550 Strix-F bricked itself after a bios update.


two 1440p 280Hz Displays on a 7900XTX at the desktop. (this would cost me 260€ per year while it does literally nothing all day long)

View attachment 338156

I guess it makes more sense you would higher failure rate considering you use so much hardware. I never thought of it that way.

Yeah, one reason I like my Steam Deck OLED, lot of games I play like indie games or older games, I am only running 4 or 7 watts all day long to game. Steam Deck 2 is going to be a must buy for me, just for the sheer electricity savings I will have.

 

[Zach_01]

The most recent idle power usage change as I understand it is as a result of the newest AMD driver. Ancient Gameplay guy (Fabio Pisco I think his name is) did a YouTube video on it.

I am not saying all 7900XTX models have the same temperate observations, I am saying this particular Powercolor 7900XTX Liquid Devil model is intended to work this way. It works and works well but with a very high delta to the GCD hotspot. The behaviour is consistent since I got the card in September last year in a variety of games: X4 Foundations, Monster Hunter World, Remnant 2, Outer Worlds, Starfield, Cyberpunk, Baldur's Gate, Days Gone etc. Depending on the game, the hit on the CPU is different but the delta on the GPU temperature remains throughout.

As I said in my earlier post, a card that takes in 550W is going to get hot. This is the default bios and default behaviour for the card. I note your card pulls in 100W less. I'll run Superposition at 4K optimised for you anyway (see the attached screenshot). The ambient temperature is 22.5C and 46% humidity. The score was 22,985 btw. This is not indicative of gaming temperatures and the delta is greater when gaming for extended periods (more than 1 hour).

I communicated with my retailer about an RMA prior to commencing the two repastes and was told that the RMA was unlikely to succeed as the card was in spec (and I it is to be fair). I even engaged directly with Powercolor to see whether this is indeed the expected behaviour, but they just don't/can't give a straight answer. This is what I get back after several tries:

Dear User,
Please ensure there is enough water flow through the RX7900XTX 24G-W/OC, and the water temperature is kept as low as possible.
If you are still worried about the abnormal temperature of the graphics card, we recommend that you return to the original place of purchase to verify.
Thanks for your inquiry from PowerColor website.

Actually 4K optimized in superposition is less intensive from lower res with extreme settings but I’ll try the same when I get back home from work.
I saw the results… you have 55+C max delta at ~390W TBP.
Even if hotspot is well within specs it doesn’t look right to me at all.

EDIT:
typo

 

[Space Lynx]

Even if hotspot is well within specs it does look right to me at all.

you mean doesn't ?

based on your sentence structure, I think you mean doesn't not does.

 

[Zach_01]

you mean doesn't ?

based on your sentence structure, I think you mean doesn't not does.

Thank you!

 

[LabRat 891]

The most recent idle power usage change as I understand it is as a result of the newest AMD driver. Ancient Gameplay guy (Fabio Pisco I think his name is) did a YouTube video on it.

This video?

 

[mama]

This video?

No, not this video. He reviews all new drivers as they come out and my recollection is that for some inexplicable reason he noted idle power draw increase when he updated to a newer driver (like AusWolf above). It think it was a couple of weeks ago but my impression of time passed and actual time isn't always the same. It could be from a couple of months ago.

 

[LabRat 891]

No, not this video. He reviews all new drivers as they come out and my recollection is that for some inexplicable reason he noted idle power draw increase when he updated to a newer driver (like AusWolf above). It think it was a couple of weeks ago but my impression of time passed and actual time isn't always the same. It could be from a couple of months ago.

Regardless, this thread and Ancient's content will be helpful to me, come Monday (hopefully).
Hoping the Sapphire Nitro 7900 doesn't have screws falling out of it, or worse.

Seems like everyone's QC and build quality has gone to hades.

 

[Zach_01]

The most recent idle power usage change as I understand it is as a result of the newest AMD driver. Ancient Gameplay guy (Fabio Pisco I think his name is) did a YouTube video on it.

I am not saying all 7900XTX models have the same temperate observations, I am saying this particular Powercolor 7900XTX Liquid Devil model is intended to work this way. It works and works well but with a very high delta to the GCD hotspot. The behaviour is consistent since I got the card in September last year in a variety of games: X4 Foundations, Monster Hunter World, Remnant 2, Outer Worlds, Starfield, Cyberpunk, Baldur's Gate, Days Gone etc. Depending on the game, the hit on the CPU is different but the delta on the GPU temperature remains throughout.

As I said in my earlier post, a card that takes in 550W is going to get hot. This is the default bios and default behaviour for the card. I note your card pulls in 100W less. I'll run Superposition at 4K optimised for you anyway (see the attached screenshot). The ambient temperature is 22.5C and 46% humidity. The score was 22,985 btw. This is not indicative of gaming temperatures and the delta is greater when gaming for extended periods (more than 1 hour).

I communicated with my retailer about an RMA prior to commencing the two repastes and was told that the RMA was unlikely to succeed as the card was in spec (and I it is to be fair). I even engaged directly with Powercolor to see whether this is indeed the expected behaviour, but they just don't/can't give a straight answer. This is what I get back after several tries:

Dear User,
Please ensure there is enough water flow through the RX7900XTX 24G-W/OC, and the water temperature is kept as low as possible.
If you are still worried about the abnormal temperature of the graphics card, we recommend that you return to the original place of purchase to verify.
Thanks for your inquiry from PowerColor website.

407W limit from adrenalin
Vcore max: 1050mV from adrenalin
GPU max edge temp: 52C
GPU max hotspot temp: 68C (max delta: 16C)

On a liquid cooled GPU at 390W a 55+C delta would be unacceptable to me
But thats me doing me

 

[AusWolf]

I wouldn’t sell the card “just” for that. Personally I don’t see it as much of an issue. Temperatures are low still even with 30W idle. Other than this, the nitro+ is great card and I’m happy with it. And if this idle power goes down again in future, then it’s +1.

Below 30 W is not a big problem, but I'd say below 20 W would be ideal. At 50 W, the card's fans have to keep turning on and off periodically to keep the GPU at around 50 °C. It's extra wear on the fan motors, not to mention electricity isn't cheap in the UK. I don't like the idea of sleeping while my GPU alone eats more power than a light bulb while doing nothing (sometimes I leave the PC on to download stuff). It's like leaving the living room lights on for the night for no reason.

No wonder I run my system RAM at 4800 MHz while it could do 6000. It saves me 10 W on the CPU's idle power while having absolutely no effect on system performance.

I have to say, this chiplet design might be great for AMD with the cost savings, but not so much for us.

 

[Zach_01]

Below 30 W is not a big problem, but I'd say below 20 W would be ideal. At 50 W, the card's fans have to keep turning on and off periodically to keep the GPU at around 50 °C. It's extra wear on the fan motors, not to mention electricity isn't cheap in the UK. I don't like the idea of sleeping while my GPU alone eats more power than a light bulb while doing nothing (sometimes I leave the PC on to download stuff). It's like leaving the living room lights on for the night for no reason.

No wonder I run my system RAM at 4800 MHz while it could do 6000. It saves me 10 W on the CPU's idle power while having absolutely no effect on system performance.

I have to say, this chiplet design might be great for AMD with the cost savings, but not so much for us.

Ahh
Didnt think of that actually as Nitro+ 7900XTX has a massive cooler that can keep hotspot temp below 45C, and fans start around 60~62C.
Even at 80-85W temp hovers around 50~55C. When I was playing FarCry6 and going into map and/or game menus like gear and stuff fans stop in a few secs.
Needs then around 100W if not more to kick fans on at 62C

 

[AusWolf]

This video?

I've just tried this. It only works temporarily, unfortunately. When you activate it, it's great, but once I quit a game, idle power is back to 30 W.

Besides, messing with the refresh rate makes me lose VRR (FreeSync) for some reason.

 

[Zach_01]

I've just tried this. It only works temporarily, unfortunately. When you activate it, it's great, but once I quit a game, idle power is back to 30 W.

Besides, messing with the refresh rate makes me lose VRR (FreeSync) for some reason.

Cant remember a detail from video... do I need to restart after making the custom res?
I tried custom 142, 143, 145 and does not accept it, turns back to previous settings.
My always setting is 144
I tried 60, 100, 120, 175 with no power drop, without restart though.

Just to clarify,
with 28~30W I have <45C
48-50W: ~50C
~55W: 53C
80~85W: 55~60C

 

[AusWolf]

Cant remember a detail from video... do I need to restart after making the custom res?
I tried custom 142, 143, 145 and does not accept it, turns back to previous settings.
My always setting is 144
I tried 60, 100, 120, 175 with no power drop, without restart though.

Just to clarify,
with 28~30W I have <45C
48-50W: ~50C
~55W: 53C
80~85W: 55~60C

I just had a look... with the monitor off, board power is around 8-9 W, which is awesome. Now, with the monitor back on, VRAM clock jumps straight to 909 MHz, and power consumption to 45 W. It's funny that it consumes less power as I'm typing this comment in Chrome than it does doing nothing on the Windows desktop, as Chrome lets the VRAM clock fluctuate a bit, while it's always 909 MHz on the desktop.

It was fine for a long time. I really don't understand what happened with these 24.x.x driver versions. I also don't get why reverting to an older version doesn't work.

Update:

I had enough of the issue, and put my 6500 XT into the system. It was fine for a while, but then I noticed that the VRAM clock was stuck at 1057 MHz during idle. I don't know by how much that bumped the idle power up, as RDNA 2 only reports GPU chip power draw, not the total board power. So then, I thought, whatever, I'll just live with this, and put the 7800 XT back. And now, it's fine(-ish). VRAM clock goes back into the 90-100 MHz range at idle, and power is at 24 W. What... the... hell?

 

[mama]

407W limit from adrenalin
Vcore max: 1050mV from adrenalin
GPU max edge temp: 52C
GPU max hotspot temp: 68C (max delta: 16C)

On a liquid cooled GPU at 390W a 55+C delta would be unacceptable to me
But thats me doing me

View attachment 338216

Thanks for this. Well, comparing the results and observations on your card to mine it looks like I pickup a lemon! Lucky me.

 

[Zach_01]

Thanks for this. Well, comparing the results and observations on your card to mine it looks like I pickup a lemon! Lucky me.

Not sure what could be wrong… and even if is something wrong in the end.
Have ever tried to tighten block around the GPU die?
Not put too much force tho and break anything…

I can see your vram temp is exceptional which indicates that the water cooler is working properly.

Or maybe vendor is a bit too much aggressive on voltage transitions/bursts. Have you tried to undervolt it?
1100mV would be a good start and then work it down.

 

[LabRat 891]

Count me in





Thankfully, this one isn't some (re-)assembly required

 

[AusWolf]

Count me in

View attachment 338512

View attachment 338513View attachment 338514

Thankfully, this one isn't some (re-)assembly required

Nice. ConGREtulations!

 

[tpa-pr]

Count me in

View attachment 338512

View attachment 338513View attachment 338514

Thankfully, this one isn't some (re-)assembly required

Congratulations! Looking forward to some benchmarks. I'm slowly but surely selling the majority of my friends on AMD hardware and this could be another option to sell.