Is RX 9070 VRAM temperature regular value or hotspot?

[JustBenching]

I didn't buy a 9000 series for that reason. The 7800X3D is tweaked such that it never reaches >85C.

Hot CPUs are a no no. I make my systems last. >10 years is standard. 15 happens. Still running a 3570k rig downstairs as HTPC, its pretty cost effective at this point Haven't changed anything in it except a PSU. 25 dollar cooler on it.

I can tell you some actual real world facts about vram temps. My 3090 - brand new, was hitting Top 10 and top 5 records on multiple benchmarks (indigo, supo, time spy, they are still up there you can check them). I was lucky enough and got a card that could do +1700 to 2000 on the memory. 1 year later, after my vram was running at 90-94C (again, was lucky enough to get a good card, cause a lot of 3090s were hitting over 100!!) I couldn't replicate these scores anymore, my vram was throwing artifacts at anywhere near those clocks. It was very obviously very degraded. If I was just running stock all the time I wouldn't have noticed obviously.

 

[Vayra86]

Mine is bone stock and still doesn't reach 85 ˚C (83-84 max under full load).


I like that sentiment very much.

Yeah I just use pbo.

I can tell you some actual real world facts about vram temps. My 3090 - brand new, was hitting Top 10 and top 5 records on multiple benchmarks (indigo, supo, time spy, they are still up there you can check them). I was lucky enough and got a card that could do +1700 to 2000 on the memory. 1 year later, after my vram was running at 90-94C (again, was lucky enough to get a good card, cause a lot of 3090s were hitting over 100!!) I couldn't replicate these scores anymore, my vram was throwing artifacts at anywhere near those clocks. It was very obviously very degraded. If I was just running stock all the time I wouldn't have noticed obviously.

I've degraded the 3570K as well in the past. It runs conservatively now, more like an underclock with a TDP limit and still a turbo set +200mhz over stock. It cán stretch its legs, but won't be exceeding 80W doing so. It used to do 4.6 at 80W, now it'll do 4.4 at best on a single core.

 

[AusWolf]

I can tell you some actual real world facts about vram temps. My 3090 - brand new, was hitting Top 10 and top 5 records on multiple benchmarks (indigo, supo, time spy, they are still up there you can check them). I was lucky enough and got a card that could do +1700 to 2000 on the memory. 1 year later, after my vram was running at 90-94C (again, was lucky enough to get a good card, cause a lot of 3090s were hitting over 100!!) I couldn't replicate these scores anymore, my vram was throwing artifacts at anywhere near those clocks. It was very obviously very degraded. If I was just running stock all the time I wouldn't have noticed obviously.

Maybe it was not meant to do +2000?

The 9070 XT got 20 Gbps chips that run exactly at 20 Gbps, so I'm not gonna touch them on my card, ever, that's 100% certain.

 

[JustBenching]

Yeah I just use pbo.


I've degraded the 3570K as well in the past. It runs conservatively now, more like an underclock with a TDP limit and still a turbo set +200mhz over stock. It cán stretch its legs, but won't be exceeding 80W doing so. It used to do 4.6 at 80W, now it'll do 4.4 at best on a single core.

My 12900k is messed up as well yeah, but in general it's harder to degrade cpus, those care mostly about voltages. Vram on the other hand can very easily get cooked by temps.

Maybe it was not meant to do +2000?

The 9070 XT got 20 Gbps chips that run exactly at 20 Gbps, so I'm not gonna touch them on my card, ever, that's 100% certain.

It was not indeed, but I only did that for benchmarks, not when running 24/7. The degradation happened from the temps, clockspeeds don't cause any degradation (unless you add voltage of course, but in that case it's the voltage that causes it.). But it didn't stop there mind you, 2nd year of ownership one (or, at least one) of the vram chips couldn't even do default clocks at the out of the box 90-94c. I had to change the pads and drop the temps to 80c or below to get stability back. The card is still working fine almost 5 years in but I don't think that would have been the case without the pad swap

 

[AusWolf]

It was not indeed, but I only did that for benchmarks, not when running 24/7. The degradation happened from the temps, clockspeeds don't cause any degradation (unless you add voltage of course, but in that case it's the voltage that causes it.). But it didn't stop there mind you, 2nd year of ownership one (or, at least one) of the vram chips couldn't even do default clocks at the out of the box 90-94c. I had to change the pads and drop the temps to 80c or below to get stability back. The card is still working fine almost 5 years in but I don't think that would have been the case without the pad swap

Couldn't you just RMA it? It shouldn't have degraded within 2 years of normal use.

 

[JustBenching]

Couldn't you just RMA it? It shouldn't have degraded within 2 years of normal use.

No, I had opened it, deshrouded and lm. Out of warranty in EU

 

[TSiAhmat]

Regarding the vram, 30xxx gpus (especially the 3090 with vram on the back) could hit up to 110c. Nvidia was saying it's normal, but it isn't. Use them enough and they will 100% degrade. I know, I've been there.

How hot did your 3090s Vram get in it's prime? Up to 110 C°?

 

[rusty caterpillar]

Couldn't you just RMA it? It shouldn't have degraded within 2 years of normal use.

Even without opening or deshrouding the card, RMA is a waste of time. From straight refusal to core reflow which is bad idea. Asus, MSI, Gigabyte, PNY, EVGA.

 

[freeagent]

o acting like it's normal or 95c is the target doesn't really check out.

I don't let my CPU run at 95, people with weak cooling do. BTW its only 5c warmer then Zen 3's 90c cap. Wait, the 5600X can take 95 too.

 

[TechBuyingHavoc]

How has it degraded if it's still running the same clocks as 3 years ago?

Try undervolting the card until it is barely stable. Try again in 3 years. You will see the degradation (it will take more voltage to achieve a given clock speed).

 

[Vayra86]

Even without opening or deshrouding the card, RMA is a waste of time. From straight refusal to core reflow which is bad idea. Asus, MSI, Gigabyte, PNY, EVGA.

Wha? Of course not. You just have to have valid grounds for RMA, and take pictures of said product before sending. Create a paper trail and don't dick around, keep it civil and just exercise your rights as a consumer.

The only reason RMA is shit is because people don't persevere and companies discovered they can just bullshit you. Its up to consumers to do better. Zero issues here with RMAs I've done.

 

[AVATARAT]

Wha? Of course not. You just have to have valid grounds for RMA, and take pictures of said product before sending. Create a paper trail and don't dick around, keep it civil and just exercise your rights as a consumer.

The only reason RMA is shit is because people don't persevere and companies discovered they can just bullshit you. Its up to consumers to do better. Zero issues here with RMAs I've done.

RMA for high temperatures is a waste of time in most situations.
I RMA'd my 7900 XTX because the hotspot paste was badly baked (it was getting 100C on undervoltage, and had reached 110C on stock. Asrock just returned me another card with a less baked hotspot...
So if the owner wants everything to be ok, he should open the card and change the paste/pads and yes, the warranty will be void.

 

[Vayra86]

RMA for high temperatures is a waste of time in most situations.
I RMA'd my 7900 XTX because the hotspot paste was badly baked (it was getting 100C on undervoltage, and had reached 110C on stock. Asrock just returned me another card with a less baked hotspot...
So if the owner wants everything to be ok, he should open the card and change the paste/pads and yes, the warranty will be void.

And you accepted that? Do better.

There is a very simple bottom line. You paid for a new product. You must get a new product that works exactly as advertised. Non negotiable, otherwise our contract is null and void and I need a full refund (or a roundtrip on social media and press to highlight this discourse).

This is how consumers should be dealing with this, honestly. No fckin mercy. Polite, but decisive.

 

[AVATARAT]

And you accepted that? Do better.

There is a very simple bottom line. You paid for a new product. You must get a new product that works exactly as advertised. Non negotiable, otherwise our contract is null and void and I need a full refund (or a roundtrip on social media and press to highlight this discourse).

This is how consumers should be dealing with this, honestly. No fckin mercy.

It was my old card that I had bought second hand, but it has a warranty. And I was 3 weeks without the card and was lucky it was "fast".
Manufacturers in most cases don't care about high temperatures unless there is something wrong, like in my situation when the card throttled.

 

[Vayra86]

It was my old card that I had bought second hand, but it has a warranty. And I was 3 weeks without the card and was lucky it was "fast".
Manufacturers in most cases don't care about high temperatures unless there is something wrong, like in my situation when the card throttled.

Yeah just 'high temps' isn't grounds for RMA, obviously. Its again brutally simple... it either works within spec or it doesn't.

 

[AVATARAT]

Yeah just 'high temps' isn't grounds for RMA, obviously. Its again brutally simple... it either works within spec or it doesn't.

High temperatures will just reduce the life of the device because the degradation is faster, but not enough to be under warranty, and they don't care what happens after that.

 

[Vayra86]

High temperatures will just reduce the life of the device because the degradation is faster, but not enough to be under warranty, and they don't care what happens after that.

Yeah that is probably how I should read that statement from @rusty caterpillar . Thanks

 

[JustBenching]

How hot did your 3090s Vram get in it's prime? Up to 110 C°?

90 to 94c, I was one of the lucky ones.

 

[rusty caterpillar]

So if the owner wants everything to be ok, he should open the card and change the paste/pads and yes, the warranty will be void.

And only a small % of users will do that and prefer to live in denial. Some of them not observing the degradation and will just upgrade faster their GPU blaming something else rather the high temps which was "officially supported"

Mine is too..

Till you remove at least back plate and change pads with thermal putty and add more, covering wider areas just stick a 92 mm fan or 120(only if your card is 14 cm wide) in pull on the back plate aprox @ 1000 rpm, also increase rear case fan RPM.
I also cut the grill, way too obstructing in my case.

Decrease your GPU temps with a single fan

Edit: use the rubber mounting pins to set the fan height on the card more than it shows in the video, at least 10mm height. Is more efficient.
If you have issues like me with trapped hot air coming form GPU heatsink exhaust, trapped in between the card and side panel, angle the fan or use 120 fan and move slightly over the edge of the card.

Targeted airflow always worked in my experience. Just in very few instances airflow pressure inside a case is so powerful to efficiently cool a hot backplate.
If a setup as above doesn't work or is just lowering 1 C, than we have:
1.Poor quality pads
2.Pads are covering very small area, not enough for back plate to really take all the heat traversing PCB(fast enough) and is acting more as a heat trap.
3. Plastic film inside the backplate, not the one you have to peel, please, the other one.
4.Lack of pads
5. Secondary plate without thermal links in between > Sapphire

 

[TSiAhmat]

90 to 94c, I was one of the lucky ones.

doesn't sound to "high" (compared to what I've read online)

also what I just noticed the 5090 FE vram runs pretty hot again (94C°)



Well only time will tell...

I'll report if anything is amiss with my card.

 

[rusty caterpillar]

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.

Oh, are refined my friend but, for efficient and insidious planned obsolesces

 

[Zach_01]

My 9070 non-xt is a 245W model and max temp for VRAM is 82°C. When GPU is almost idle, VRAM temp is already at 62°C while my GPU temp is 40°C and hotspot is 42°C

1. What is the idle power (TBP) consumption of the card?
2. What is the idle speed of VRAM?
3. What is the idle speed of GPU?

 

[NSR]

1. What is the idle power (TBP) consumption of the card?
2. What is the idle speed of VRAM?
3. What is the idle speed of GPU?

I'm not the guy you quoted but I also have a 240W 9070 and I get the same numbers on my memory temps.

1. Idle power consumption - 25W
2. Idle speed of Memory Cock - 900MHz
3. Idle speed of GPU Clock - 50MHz to 100 MHz

GPU is enjoying life at 40-45 degrees C including the hot spot.
Memory is hanging around at 60-65 degrees C.

 

[TSiAhmat]

I'm not the guy you quoted but I also have a 240W 9070 and I get the same numbers on my memory temps.

1. Idle power consumption - 25W
2. Idle speed of Memory Cock - 900MHz
3. Idle speed of GPU Clock - 50MHz to 100 MHz

GPU is enjoying life at 40-45 degrees C including the hot spot.
Memory is hanging around at 60-65 degrees C.
View attachment 391712View attachment 391713

weird, my idle was around 8 ~ 14 W... Have to check it again when I am at home.

 

[rusty caterpillar]

weird, my idle was around 8 ~ 14 W... Have to check it again when I am at home.

Yes, idle, should be around 10 Watt, memory clock to high for idle IMO.

I'm not the guy you quoted but I also have a 240W 9070 and I get the same numbers on my memory temps.

1. Idle power consumption - 25W
2. Idle speed of Memory Cock - 900MHz
3. Idle speed of GPU Clock - 50MHz to 100 MHz

GPU is enjoying life at 40-45 degrees C including the hot spot.
Memory is hanging around at 60-65 degrees C.
View attachment 391712View attachment 391713

You have 2 screens linked to your card?