Is RX 9070 VRAM temperature regular value or hotspot?

[rusty caterpillar]

With a car's engine temp, you know where the redline is. Do you know where it is on GDDR6 VRAM chips?

Yes we know, GDDR6 was used on Nvidia 2000, 3000, 4000 and some AMD cards to. On NV died before mentioned cards without hitting thresholds for throttling and they run on lower temps than they run now on 9070XT. I say again, I hope is a different revision of GDDR6 and works on different parameters but, it just might not be.

Nobody said you should leave in fear of something you don't know. Observe, question, investigate don't hit the official info provided wall, and call it a day. Gave up and open the wallet is what they want from us in the end of the day.

It seems also ignore the videos I gave you, but I would say is plenty of data we only have atm, from somebody repairing 2-5 X GPU/day.

Micron chips starting with 8 manufactured in 2018. Replacing mem modules on Turing cards EVGA and Gigabyte >>>> GDDR6

Replacing mem modules on 1080 Ti >>>> GDDR5X

Replacing mem modules and title goes like "Common problem with most GPUs" >>>>> GDDR6X 3080 Ti

Is a good start to familiarize your self with several issue of now days GPUs

Who makes the coolest running Radeons?

I was under the impression that they all ran hot.

Like 90-100c is normal, as crazy as that sounds.

Numbers like that are only normal to Radeon users.

When I had HD 7950 3GB was a very cool card and it was Sapphire Nitro, for 9000 IDK we have to wait for somebody to replace stock pads and add more pads on the back plate + remove plastic film if there, not all AIB added the plastic film.

What do you base that on?

Because previous cards using GDDR6 didn't had that temps on VRAM, still VRAM fried on lower temps. But why do you think 92-94 C is normal on VRAM? because AIB said so. Who is making money when you replacing your bricked card, me or AIB?

They were wrong, and then it got fixed. Now, only the Nvidia FE 5090 has such issues (supposedly because of the angled placement).

Power gate following the connector is insufficient and weak . Is not only the cable or connector. Was the same on 4090 so they didn't corrected anything they just ported the issue like nothing happened to 5000 series. AIB followed the bad design on 4090s

NOICE power gate


Only fools repeat mistakes or those who want to make money in an efficient way, "no need to design another flawed PCB we have one"

Where do you have room on this silly cracker they call PCB, to add enough power rails and fuses. Is just good for dunking in Ovaltine and take a bite. 4090 FE
Did I mention heat by proximity on such crammed power gate?

 

[Zach_01]

Do we know which of these cards have backplate pads for additional VRAM cooling?

Also look at the darn rpm…
Do we really expecting nice cooling with 1200rpm at 370W (Taichi)?
Or 900rpm at 325W?

Has any 9070XT owner tried to set a bit more aggressive fan curve? Like 1600~1800rpm?

I don’t expect to properly cool a 330~350+W card with farts, because it’s like that at 1000~1400rpm.

 

[tugrul_LordOfDrinks]

We need an isolated cooling system for the memory. At 30Celcius that memory could have 2 tiers better timings.

 

[droopyRO]

Do we really expecting nice cooling with 1200rpm at 370W (Taichi)?
Or 900rpm at 325W?

This is the issue with lower end cards too, i had both a XFX dual cooler and a Sapphire Pulse RX 6650 XT. Both barely spun their fans at 1000-1200 rpm and memory temps got to about 90-94ºC. Just bumping it up to about 1600-1700 solved the issue by dropping the vRAM temperature in the 80s.

 

[Event Horizon]

On my 9070 Prime, at stock, the fans won't even turn on until the GPU core temp reaches 55c, and even then it usually goes up to 500rpm. AMD Adrenaline does not support adjusting the zero fan threshold, which is a bummer. There's the option to disable it entirely, but I'm not sure if I want that. If they're worried about pebcaks frying their cards they could make it a setting that only adjusts downwards (example 55c -> 45c).

 

[Vayra86]

This is the issue with lower end cards too, i had both a XFX dual cooler and a Sapphire Pulse RX 6650 XT. Both barely spun their fans at 1000-1200 rpm and memory temps got to about 90-94ºC. Just bumping it up to about 1600-1700 solved the issue by dropping the vRAM temperature in the 80s.

Well yeah but look at the GPU core.

It just wouldn't quite make sense to cool a hotspot of 82C further down. I'd rather think, as a user, I'd have heatroom for more powah - OR I could run my fans even slower for less noise.

 

[rusty caterpillar]

View attachment 391292

Do we know which of these cards have backplate pads for additional VRAM cooling?

Also look at the darn rpm…
Do we really expecting nice cooling with 1200rpm at 370W (Taichi)?
Or 900rpm at 325W?

Has any 9070XT owner tried to set a bit more aggressive fan curve? Like 1600~1800rpm?

I don’t expect to properly cool a 330~350+W card with farts, because it’s like that at 1000~1400rpm.

Depends how many courageous 9000 users will be take apart their cards, at least the backplates.

 

[Knight47]

View attachment 391292

Do we know which of these cards have backplate pads for additional VRAM cooling?

Also look at the darn rpm…
Do we really expecting nice cooling with 1200rpm at 370W (Taichi)?
Or 900rpm at 325W?

Has any 9070XT owner tried to set a bit more aggressive fan curve? Like 1600~1800rpm?

I don’t expect to properly cool a 330~350+W card with farts, because it’s like that at 1000~1400rpm.

Only XFX did not use backplate pads for the VRAM.

On my 9070 Prime, at stock, the fans won't even turn on until the GPU core temp reaches 55c, and even then it usually goes up to 500rpm. AMD Adrenaline does not support adjusting the zero fan threshold, which is a bummer. There's the option to disable it entirely, but I'm not sure if I want that. If they're worried about pebcaks frying their cards they could make it a setting that only adjusts downwards (example 55c -> 45c).

Does the Prime ramp up the fans to 2000+rpm first before going down to 500rpm like on nvidia?

 

[Zach_01]

Well yeah but look at the GPU core.

It just wouldn't quite make sense to cool a hotspot of 82C further down. I'd rather think, as a user, I'd have heatroom for more powah - OR I could run my fans even slower for less noise.

But this is the case.
In order to keep VRAM at 75-80C the GPU hotspot of my 7900XTX Nitro+ stays around 70~75C. And this thing has a 3.5 slot cooler with pads under the metal backplate.
At 382W TBP the fans are averaging at ~1900rpm (3200max) for the mentioned thermals above. I can hear the airflow but not the fans. Mind you I have no case and the GPU is 2 feet away from my left ear. It’s not an annoying sound.

On my 9070 Prime, at stock, the fans won't even turn on until the GPU core temp reaches 55c, and even then it usually goes up to 500rpm. AMD Adrenaline does not support adjusting the zero fan threshold, which is a bummer. There's the option to disable it entirely, but I'm not sure if I want that. If they're worried about pebcaks frying their cards they could make it a setting that only adjusts downwards (example 55c -> 45c).

I do t see problem. My XTX with zero fan ON start moving fans at 60+C GPU hotspot, and as I said above after a while GPU hotspot avg at 70~75C.
If it’s going to be stable at 70-75C eventually, why do I need the fans to start below 60-65C?

 

[Event Horizon]

Does the Prime ramp up the fans to 2000+rpm first before going down to 500rpm like on nvidia?

I haven't noticed anything like that happening. Does this ramping up last long enough to be seen in the performance overlay, or do I need to put my ear to the case?

 

[AusWolf]

There are bandwidth test programs to measure bandwidth. OpenCL based programs work for all cards including AMD. CUDA preferred for Nvidia. I use them to find best mem oc, not blindly apply +2000.

Do you even trust the temperature sensor? What if it reports 20Celcius lower than the actual temperature? 95C => 115C doomed to fail quick.

At least have some safety margin in there, like 80C so the worst can be 100C in real. Thats why people have questions.

I generally prefer 10C delta safety margin against measurement error. 95C is right at the edge of danger zone. I would have questions if my mem had this issue.

I don't have a reason to trust the sensor, but I also don't have a reason not to trust it.

If the card fails, I'm gonna have lots of questions. Until then, there's nothing to do.

I use this one memtest_vulkan for errors/bandwidth, but beware, it generates a lot of heat.

Usually a short test is enough.

I can imagine if it only stresses VRAM, then the fans won't spin up as they're tied to core temperature. This is not a realistic load, so I think I'll pass on this one. I don't use furmark, either.

Who makes the coolest running Radeons?

I was under the impression that they all ran hot.

Like 90-100c is normal, as crazy as that sounds.

Numbers like that are only normal to Radeon users.

We're only talking about the VRAM here. The core is very cool even on my PC Reaper. 60 °C edge, 85 °C hotspot.

 

[freeagent]

We're only talking about the VRAM here.

I was too

Mostly.. I just wonder why their VRAM runs so warm on this gen. I figured these would have been some highly refined machines.

 

[droopyRO]

Well yeah but look at the GPU core.

Yup. My 6650 XT goes to about 78ºC on the core and 93ºC on the memory at 1200 rpm. This 9070 cards are in their 60s for core and high 80s low 90s for memory. This is the weird part for me :shrug:

 

[Event Horizon]

I was too

Mostly.. I just wonder why their VRAM runs so warm on this gen. I figured these would have been some highly refined machines.

Second hand heat from the core? Could explain why the VRAM is much hotter on the XTs compared to the non-XTs despite having the same VRAM bandwidth and capacity. Could also be that the XT's higher number of shading units and higher clocks makes it more memory hungry.

 

[freeagent]

Second hand heat from the core? Could explain why the VRAM is much hotter on the XTs compared to the non-XTs despite having the same VRAM bandwidth and capacity. Could also be that the XT's higher number of shading units and higher clocks makes it more memory hungry.

I am comparing to Nvidia because that is all I know..

 

[Event Horizon]

I am comparing to Nvidia because that is all I know..

My old 3060ti also used GDDR6, but clocked much lower, and with a more aggressive fan curve. This is GDDR6 pushed to near its limit.

 

[freeagent]

This is GDDR6 pushed to near its limit

Mine is too..

 

[AusWolf]

I was too

Mostly.. I just wonder why their VRAM runs so warm on this gen. I figured these would have been some highly refined machines.

Perhaps they are. Perhaps it's just normal, just like 95 °C on Ryzen 7000/9000 is normal. What's weird for me is that no one seems to be considering that possibility here.

 

[tugrul_LordOfDrinks]

10 years later:
(AMD) "it is normal to have lava flowing down the contact point".
(Nvidia) "RTX10080 = 14000 cores, RTX10090 = 56000 cores, with 2x6 connector + cable cooling"

 

[AusWolf]

10 years later:
(AMD) "it is normal to have lava flowing down the contact point".
(Nvidia) "RTX10080 = 14000 cores, RTX10090 = 56000 cores, with 2x6 connector + cable cooling"

Don't get me wrong, but we've had this argument with the introduction of the hotspot sensor on RX 5000, and also during the Ryzen 7000 launch. I haven't heard of any of them fail due to high temperatures.

 

[tugrul_LordOfDrinks]

Don't get me wrong, but we've had this argument with the introduction of the hotspot sensor on RX 5000, and also during the Ryzen 7000 launch. I haven't heard of any of them fail due to high temperatures.

Amd says its ok to run CPU at 95. But not for vram. When amd says same for VRAM, its ok. But amd is not producing vram. So we have to listen to samsung, micron, hynix. What they say?

 

[freeagent]

just like 95 °C on Ryzen 7000/9000 is normal.

That is normal at absolute full load, playing a game is easy. But for Radeon I see the high temps as excessive, normal or not..

Anywhoo, I will continue to watch those repair guys on the toob. Always interesting stuff.

 

[AVATARAT]

I can imagine if it only stresses VRAM, then the fans won't spin up as they're tied to core temperature. This is not a realistic load, so I think I'll pass on this one. I don't use furmark, either.

100% case fans, -150 voltage offset, and you can see the result on the right, of course with less offset the temperature will be more, but I test VRAM so I can reduce GPU voltage.

 

[AusWolf]

Amd says its ok to run CPU at 95. But not for vram. When amd says same for VRAM, its ok. But amd is not producing vram. So we have to listen to samsung, micron, hynix. What they say?

Micron says 95 °C max on edge temp is safe. I don't think they've said anything about the hotspot. With a 90 °C hotspot, your edge temp is obviously a lot lower.

 

[freeagent]

100% case fans, -150 voltage offset, and you can see the result on the right, of course with less offset the temperature will be more, but I test VRAM so I can reduce GPU voltage.

View attachment 391371

See, that is what I wanted to see. I haven't seen this program before.. my mem temp hit 72c and I let it run 3 extra iterations compared to yours. My fans spun up to 1284rpm, but my case fans are at full speed.

Edit:

I see nothing wrong with that..