AMD RX 7000 series GPU Owners' Club

[Kabouter Plop]

did it lower your temps any? my 7900 xt has so much metal on it, that i am considering liquid metal. liquid metal doesn't make sense for most cards true, but mine is a giant card with all metal... so it might act more like a laptop... not sure if i want to risk liquid metal or not yet though. also i don't know if my baseplate on the heatsink is copper or what

My hotspot was 95-99c at 470w total board power with PTM7950 it was 72c hotspot at same total board power in a game that runs much lower then some games so im sure i could hit even 110c hotspot at stock.

at 410w total board power it was about 92c max hotspot on both stock paste and MX-4 after 2 weeks stock paste got worse after like 3 months altho it already hit 90c hotspot after 1 month it just ended up doing that even at really low water temps.

I would take PTM7950 before liquid metal if still not satisfied try liquid metal but i cant help with that, linustechtips had results where the ptm7950 on a cpu pushed more wattage on PTM7950 since the cpu they tested boost till thermal limit so lower temps results in pushing higher wattage getting more performance, then liquid metal did for them, i think they just tested directly on ihs tho not sure if it was a delided test.

PTM7950 is often used in laptops cos dies are just exposed so it works well, since its a phasechanging material that gets solid at 45c and below and melts above 45c but hotspot gets hot enough to make the pad melt suposely acts similar then liquid metal also gets more solid when its colder not sure at which temp.

Better to test and see if it helps if seen some users only get 7c improvement but they test it short term against other more expensive paste, usually after couple of weeks or months sometimes days in rare case temps get worse, that was case for me with MX-4 after 2 weeks i would be back at 92c hotspot having chilling watertemps like towards 20c did not lower the hotspot temps even edge over time went up.

I would say you can least expect a 10c drop on hotspot less if you compare to paste that been on there fresh im not sure if PTM7950 gets worse after couple of months or after 3 months i hope not its used in the industry so probably last quite a bit longer.

 

[Space Lynx]

My hotspot was 95-99c at 470w total board power with PTM7950 it was 72c hotspot at same total board power in a game that runs much lower then some games so im sure i could hit even 110c hotspot at stock.

at 410w total board power it was about 92c max hotspot on both stock paste and MX-4 after 2 weeks stock paste got worse after like 3 months altho it already hit 90c hotspot after 1 month it just ended up doing that even at really low water temps.

I would take PTM7950 before liquid metal if still not satisfied try liquid metal but i cant help with that, linustechtips had results where the ptm7950 on a cpu pushed more wattage on PTM7950 since the cpu they tested boost till thermal limit so lower temps results in pushing higher wattage getting more performance, then liquid metal did for them, i think they just tested directly on ihs tho not sure if it was a delided test.

PTM7950 is often used in laptops cos dies are just exposed so it works well, since its a phasechanging material that gets solid at 45c and below and melts above 45c but hotspot gets hot enough to make the pad melt suposely acts similar then liquid metal also gets more solid when its colder not sure at which temp.

Better to test and see if it helps if seen some users only get 7c improvement but they test it short term against other more expensive paste, usually after couple of weeks or months sometimes days in rare case temps get worse, that was case for me with MX-4 after 2 weeks i would be back at 92c hotspot having chilling watertemps like towards 20c did not lower the hotspot temps even edge over time went up.

I would say you can least expect a 10c drop on hotspot less if you compare to paste that been on there fresh im not sure if PTM7950 gets worse after couple of months or after 3 months i hope not its used in the industry so probably last quite a bit longer.

i think i am going to stay away from ptm7950, i read it only has 1000 cycles before it needs replaced, so if you go above 45 celsius and trigger its phase change 1000x it needs replaced around then...

i do turn my pc on and off maybe twice per day, even if i left it on, the default temp is below 45 celsius... so i would be triggering a phase change multiple times as i launch different games. so i'd probably need a repaste every 4 months... A) thats too pricey B) im too lazy to do it that often

 

[Kabouter Plop]

i think i am going to stay away from ptm7950, i read it only has 1000 cycles before it needs replaced, so if you go above 45 celsius and trigger its phase change 1000x it needs replaced around then...

i do turn my pc on and off maybe twice per day, even if i left it on, the default temp is below 45 celsius... so i would be triggering a phase change multiple times as i launch different games. so i'd probably need a repaste every 4 months... A) thats too pricey B) im too lazy to do it that often

It improves after 1000 cycles its suppose to last really long time, check the linustechtips video

 

[Space Lynx]

It improves after 1000 cycles its suppose to last really long time, check the linustechtips video

I must have misread, thanks. I may give this a go then. i get 54 and 75 hot spot in most demanding games, no oc, stock fan. i wonder if i can bring that down even more

 

[Kabouter Plop]

I must have misread, thanks. I may give this a go then. i get 54 and 75 hot spot in most demanding games, no oc, stock fan. i wonder if i can bring that down even more

Just give it a try if gives good temps keep an eye on it see if it stays like this if it does not you can always try something else in the future.

 

[Amir43]

Its very finicky to work with just get 80*80 biggest size pad and cut to size may take 2e try there plenty of tips online, keep us update how it works out not everyone gets big improvement, you can accidenticly rip the pad cos its very fragile after putting it on the gpu core for example happened to me on first attempt so i held pad in place more while i peeled 2e time.

The kit comes with 2 pieces at 25x25mm. Is the 7900 size larger than 25mm?

 

[Kabouter Plop]

The kit comes with 2 pieces at 25x25mm. Is the 7900 size larger than 25mm?

I bought mine from ebuy7 which was better 1 out f the 2 the other one comes from aliexpress ltt showed no data about that one acording to linus performed worse.
I bought the 80*80*0.2mm one from ebuy7 i am not sure if i am allowed to post link to it.
Linus mentioned safest option is moddiy if you want the legit vision actually PTM7950 only comes from 0.25mm thick so not sure if they rounded it down on ebuy7.

Honeywell PTM7950 0.2mm 8.5W/mK High Thermal Conductivity Phase Change Silicone Pad for CPU and GPU High-Performance Devices

Honeywell ptm7950 phase change thermal pad, designed for computers cpu & gpu, incorporating silicone grease and paste pad patch material. enhances device cooling. boost performance. - Honeywell Ptm7950, Honeywell Thermal Paste, Honeywell Ptm7950 - 20*30*0.2mm in one piece Other Computer Peripherals

ebuy7.com

My apoligies if this not allowed not trying to advertise anything.

edit: there might be a better vision then the one i linked even from legit source but this is the one i have cut to size the 80*80*0.2mm one that gave me instant good results.
At first it may perform worse then it gets better and better after like 5 minutes and then after an hour or a day in a week you may drop another 2c altho this can be margin of error still.
my aplication is 1 month old next teusday im testing daily to see if anything changed, hopefully this will least last 2 years but it probably last least lifetime of the device its on.

 

[Vya Domus]

I have my inlet/outlet reversed on my GPU block, I only realized after I finished assembling and bleeding my loop so I never bothered changing it to how it's supposed to be.

Does anyone know if it really makes a difference ? Looking at the layout of the block the water is flowing first over memory/VRM then over the GPU die the way I have it set up right now, I guess technically I should be getting a little bit better temps if it was the other way around.

 

[Kabouter Plop]

I have my inlets reversed on my GPU block, I only realized after I finished assembling and bleeding my loop so I never bothered changing it to how it's supposed to be.

Does anyone know if it really makes a difference ? Looking at the layout of the block the water is flowing first over memory/VRM then over the GPU die the way I have it set up right now, I guess technically I should be getting a little bit better temps if it was the other way around.

On my 6900 XT it made only 1-2c difference its worth chasing every few degree's if you care a lot about temps maybe even obsessed with it like me, i have double Mo-ra3 afterall soon currently just running one waiting to mount 2e one on wall need to wait for some one to do it first.
EKWB did state tho that it supports reverse flow not all blocks support it seen plenty topics where temps where to high.

 

[Vya Domus]

On my 6900 XT it made only 1-2c difference its worth chasing every few degree's if you care a lot about temps maybe even obsessed with it like me, i have double Mo-ra3 afterall soon currently just running one waiting to mount 2e one on wall need to wait for some one to do it first.

I also originally indented to have the loop order reservoir->rad->rad->GPU->CPU but ended up with reservoir->CPU->rad->GPU->rad because the case was too restrictive. I know loop order does technically make a difference albeit it small, I do care about the temperatures but I'll wait until I get a new case in a couple of months then I'll redo it properly.

 

[Kabouter Plop]

at reasonable flow rate you wont see a difference no matter which loop order, altho if always put it into GPU first or radiator and then gpu cpu always comes last, cos i do not care about CPU temps are hard to tame unless you delid and use liquid metal or i guess PTM7950 may work as well, haven't delided cpu yet so i am not sure.
PTM7950 relies on really high wattage chips that dump a lot of heat it can actually perform much worse if it does not melt from what if heard, which is funny cos i am watercooling so it gets hot enough to melt.

edit: my 20c improvement mainly comes after 2 weeks of MX-4 cos fresh it performed like 75c hotspot max then got worse to 92c so compared to other paste fresh application improvement might be less maybe only 5c on the hotspot if temps keep getting worse like it did for me it may give bigger improvement, if always had 92c hotspot at 410w total board power even after 1 month but only at certain watertemps like 27c towards 30c or when room is about 25c it gets no where near that hot anymore at 25c room temp.

Not to try and misslead people in thinking PTM7950 is good altho its really good acording to LinusTechTips and some other users, how well it performs probably depends on how hot your chip gets how much power go's thru it, if it does not reach the melting point it can reach worse temps but then again worse temps means it reaches the melting point anyway and then improves so not sure if it stays good after first couple of heat cycles.

 

[QuietBob]

Here's another popular racer with and without a 60fps cap. F1 23 maxed out in 4K + TAA (no RT). One lap in the built-in benchmark on the Singapore track in wet weather with a cycling camera:

Again, huge savings in power/heat leading to much lower temps and quieter fans. 53% lower GPU consumption on average, with 14% lower CPU power to boot

 

[mama]

I tried asking in my own thread but this seems the place to ask. I just purchased and installed a Liquid Devil. Looking at Reddit these cards some to have a significant delta between edge and hotspot temperatures. Mine does. Hotspot was hitting 100C in three hours of gaming. All other temperatures were really good which was weird to me. I have cut back the power and undervolted the card which has dropped temperatures but should I do anything further? Is it running normally (for this model)? Maybe a repaste with something like PTM7950? Or should I just RMA? I attach a screenshot of the system after 3 hours of Starfield (which came with the card).

 

[remekra]

I tried asking in my own thread but this seems the place to ask. I just purchased and installed a Liquid Devil. Looking at Reddit these cards some to have a significant delta between edge and hotspot temperatures. Mine does. Hotspot was hitting 100C in three hours of gaming. All other temperatures were really good which was weird to me. I have cut back the power and undervolted the card which has dropped temperatures but should I do anything further? Is it running normally (for this model)? Maybe a repaste with something like PTM7950? Or should I just RMA? I attach a screenshot of the system after 3 hours of Starfield (which came with the card).

If a water-cooled card is reaching 100 on hotspot while mine air cooled one does not exceed 90 even at 450W then yeah I would either RMA it or repaste it.

 

[AusWolf]

I tried asking in my own thread but this seems the place to ask. I just purchased and installed a Liquid Devil. Looking at Reddit these cards some to have a significant delta between edge and hotspot temperatures. Mine does. Hotspot was hitting 100C in three hours of gaming. All other temperatures were really good which was weird to me. I have cut back the power and undervolted the card which has dropped temperatures but should I do anything further? Is it running normally (for this model)? Maybe a repaste with something like PTM7950? Or should I just RMA? I attach a screenshot of the system after 3 hours of Starfield (which came with the card).

That's nearly a 50 °C delta between hotspot and junction temp - definitely a contact issue between the GPU and the coldplate. It's not normal at all.

 

[Space Lynx]

I tried asking in my own thread but this seems the place to ask. I just purchased and installed a Liquid Devil. Looking at Reddit these cards some to have a significant delta between edge and hotspot temperatures. Mine does. Hotspot was hitting 100C in three hours of gaming. All other temperatures were really good which was weird to me. I have cut back the power and undervolted the card which has dropped temperatures but should I do anything further? Is it running normally (for this model)? Maybe a repaste with something like PTM7950? Or should I just RMA? I attach a screenshot of the system after 3 hours of Starfield (which came with the card).

I agree with the above, I would RMA that. I can sometimes see 25-30 celsius difference between my core and hotspot, but 50 is not normal imo... and that is only in the most very demanding games. most games I only see 10-15 celsius between core and hotspot.

 

[Kabouter Plop]

I tried asking in my own thread but this seems the place to ask. I just purchased and installed a Liquid Devil. Looking at Reddit these cards some to have a significant delta between edge and hotspot temperatures. Mine does. Hotspot was hitting 100C in three hours of gaming. All other temperatures were really good which was weird to me. I have cut back the power and undervolted the card which has dropped temperatures but should I do anything further? Is it running normally (for this model)? Maybe a repaste with something like PTM7950? Or should I just RMA? I attach a screenshot of the system after 3 hours of Starfield (which came with the card).

To high if already tryit repaste try PTM7950
94c at 370w total board power more or less is insane i would expect least 20c drop with PTM7950 honeywell pad on the gpu core in your case, if you are in area that voids warranty when repasting outside US and outside EU or aren't confident in repasting yourself i would rma, i would do that anyway if you cannot get PTM7950 honeywell pad which is usually resold cos it can only be bought in large batches.
Defiantly way to hot, your card probably has pumpout issues or bad vapor chamber the later only RMA is the solution, if you can repaste without voiding warranty PTM7950 is defiantly worth trying, as regular past will likely pump out and bring you back to the same hotspot temps you have now.

 

[remekra]

To be honest I would just RMA it. Even if repaste would fix it. This is high end model of the high end card that costs a lot of money. It should not be accepted that it just comes like that reaching 100 degrees and to fix it yourself.

 

[Kabouter Plop]

To be honest I would just RMA it. Even if repaste would fix it. This is high end model of the high end card that costs a lot of money. It should not be accepted that it just comes like that reaching 100 degrees and to fix it yourself.

Repaste with regular paste have high doubts it would fix, it would be good for maybe 1-2 weeks then get worse and back to same hotspot, pumpout happens due uneven cooler surface or chip die being uneven, PTM7950 cannot pumpout but its also hard to get you get it from resellers, if card is within return perioid i would just return it get another one, my liquid devil 7900 XTX i repasted 2 times with MX-4 it was 75c hotspot then went back up to 92c hotspot after 1-2 weeks each time, pumpout is very annoying to deal with, which is why i wen't PTM7950, RMA is probably best option in this case 94c hotspot at about 370w total board power is insanely high.

 

[remekra]

Repaste with regular paste have high doubts it would fix, it would be good for maybe 1-2 weeks then get worse and back to same hotspot, pumpout happens due uneven cooler surface or chip die being uneven, PTM7950 cannot pumpout but its also hard to get you get it from resellers, if card is within return perioid i would just return it get another one, my liquid devil 7900 XTX i repasted 2 times with MX-4 it was 75c hotspot then went back up to 92c hotspot after 1-2 weeks each time, pumpout is very annoying to deal with, which is why i wen't PTM7950, RMA is probably best option in this case 94c hotspot at about 370w total board power is insanely high.

Yeah I agree. I only wonder how many people are affected by that. There seems to be a lot of those issues with XTX.

 

[Kabouter Plop]

Yeah I agree. I only wonder how many people are affected by that. There seems to be a lot of those issues with XTX.

AMD been having uneven die for a long time, it looks like it needs to be adressed with thermal paste that does not pump out, eitherway most thermal paste will pumpout is only showing good numbers during reviews, not when being used for a long time, i wish paste was reviewed on how many days it last until thermals degrade to a point that is obviously higher then what was reviewed on day 1 and by how much.

 

[remekra]

AMD been having uneven die for a long time, it looks like it needs to be adressed with thermal paste that does not pump out, eitherway most thermal paste will pumpout is only showing good numbers during reviews, not when being used for a long time, i wish paste was reviewed on how many days it last until thermals degrade to a point that is obviously higher then what was reviewed on day 1 and by how much.

Yup that's what happened with my Gigabyte Gaming OC XTX. For the first 6 months all good then temps were raising and raising.
Would be nice to have some investigation into it, on why exactly they have those problems with uneven dies.

 

[mama]

I will see about an RMA. Thanks for the advice.

 

[Kabouter Plop]

Did not realize you also where having liquid devil, honestly i hope they fix it for you but chances are it gets worse again if the stuff powercolor uses is bad, from my own experience had good temps in first month then got slightly worse 2e month then really bad 3e month, but using MX-4 as replacement is worse, mx-6 probably also just as bad as is anything else, so you may end up repasting yourself anyway if so i strongly advice PTM7950 mine is from ebuy7 but most legit vision if you want go safest route its slightly more expensive but moddiy is safest.

my hotspot at 410w total board power unleashed bios at stock no undervolt or anything is in 60-70 range.

if you hit 92c hotspot again on liquid devil at 22c room temp full open case all fans maxed out you have to repaste, you probably start seeing crazy deltas again to, its not just the hotspot that can be much hotter also the edge, i am usually max 42c edge temp unless i record in relive then hotspot go's down and edge go's down, probably cos chip is bulging when recording is done making hottest area touch cooling surface more while edge gets pushed away but could be wrong, this is something i cannot verify, but if my assumption is correct that would make pumpout even worse, the question is tho if it could even pumpout PTM7950 if i have to believe honeywell or linustechtips it probably cant, but that may still be a problem with normal thermal paste.

 

[mama]

Did not realize you also where having liquid devil, honestly i hope they fix it for you but chances are it gets worse again if the stuff powercolor uses is bad, from my own experience had good temps in first month then got slightly worse 2e month then really bad 3e month, but using MX-4 as replacement is worse, mx-6 probably also just as bad as is anything else, so you may end up repasting yourself anyway if so i strongly advice PTM7950 mine is from ebuy7 but most legit vision if you want go safest route its slightly more expensive but moddiy is safest.

my hotspot at 410w total board power unleashed bios at stock no undervolt or anything is in 60-70 range.

if you hit 92c hotspot again on liquid devil at 22c room temp full open case all fans maxed out you have to repaste, you probably start seeing crazy deltas again to, its not just the hotspot that can be much hotter also the edge, i am usually max 42c edge temp unless i record in relive then hotspot go's down and edge go's down, probably cos chip is bulging when recording is done making hottest area touch cooling surface more while edge gets pushed away but could be wrong, this is something i cannot verify, but if my assumption is correct that would make pumpout even worse, the question is tho if it could even pumpout PTM7950 if i have to believe honeywell or linustechtips it probably cant, but that may still be a problem with normal thermal paste.

My edge temperatures are fine. In fact, everything but the hotspot is not a concern and temperatures are good and this is in extended 4K gaming scenarios (Baldur's Gate and Starfield). I have just asked the vendor for an RMA but I'm really not sure. Maybe the card is just meant run that bit hotter (on hotspot) as a custom model.