Is RX 9070 VRAM temperature regular value or hotspot?

[AusWolf]

Don't you also believe that all engineers didn't know what they were doing with the design of the 12vh?

That's one company (Nvidia - no other maker seems to have issues with melting 5090s as far as I know). Engineers could have been overruled by higher powers there, which is totally believeable, imo.

But with the VRAM temps, we're talking about a dozen or so companies. Did all of them conspire to run VRAM hot on all of their cards? That's kind of tinfoil hat category there.

10 hours of gaming

If you can game for 10 hours a day, on any day, I'll take my hat off to you, sir.

 

[tugrul_LordOfDrinks]

Msi has designed lightning models in past, cooling the memory during HD7870 era. But today a 4070 ventus 2x saks on vram cooling. How much does "touching the memory" cost today? 10 newton force? 20 newton force? What kind of bargain is this?

I have an idea for the engineers:

Put heatsinks on memory modules and let air flow through them. Good idea right? I must be a genius.

 

[JustBenching]

That's one company (Nvidia - no other maker seems to have issues with melting 5090s as far as I know). Engineers could have been overruled by higher powers there, which is totally believeable, imo.

But with the VRAM temps, we're talking about a dozen or so companies. Did all of them conspire to run VRAM hot on all of their cards? That's kind of tinfoil hat category there.

It was a body of engineers that designed it not nvidia. It's engineers from Dell, ARM, IBM, Intel, AMD, Nvidia, Qualcomm and many more. They all deemed the 12vh safe. You are saying they are wrong, while at t he same time all the engineers from the AIB cannot be wrong.

 

[Vayra86]

If you can game for 10 hours a day, on any day, I'll take my hat off to you, sir.

This is Standard Operating Procedure man

 

[tugrul_LordOfDrinks]

It was a body of engineers that designed it not nvidia. It's engineers from Dell, ARM, IBM, Intel, AMD, Nvidia, Qualcomm and many more. They all deemed the 12vh safe. You are saying they are wrong, while at t he same time all the engineers from the AIB cannot be wrong.

Long story short, engineers failed much more important missions already. Memory cooling shouldnt be an exception.

Here's an extra:

Gimli Glider 1983: Engineers miscalculated fuel by mixing up pounds and kilograms, causing an airliner to run out of fuel mid-air. The pilots had to glide the plane to safety.

 

[AusWolf]

It was a body of engineers that designed it not nvidia. It's engineers from Dell, ARM, IBM, Intel, AMD, Nvidia, Qualcomm and many more. They all deemed the 12vh safe. You are saying they are wrong, while at t he same time all the engineers from the AIB cannot be wrong.

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

If every single AMD AIB is wrong with every single one of their cards (which I don't find believable one bit), then we'll get a statement soon, or at least examples of cards failing.

Back in the days, everybody told me that 105 ˚C on my 6750 XT's GPU hotspot was unsafe. The card has been running strong for the last 3 years, with no sign of failure.

Long story short, engineers failed much more important missions already. What makes you think memory cooling is an exception?

The fact that we're talking about dozens of engineers at dozens of different unrelated companies. What's the likelihood of all of them making the same mistake on the same generation of cards?

 

[Vayra86]

The fact that we're talking about dozens of engineers at dozens of different unrelated companies. What's the likelihood of all of them making the same mistake on the same generation of cards?

What's the likelihood of AMD simply giving a spec sheet that extends the usable range a little bit for these chips, so AIBs have done what they've been told? What if the focus overall for these products was elsewhere?

We saw something akin to that with the 5090 and one rare exception having some mitigation for bad power. Again... assumptions don't do you any favors here and they certainly don't support your stance. We just don't know how deep the rabbit hole goes and there is no track record to rely on either.

 

[tugrul_LordOfDrinks]

The fact that we're talking about dozens of engineers at dozens of different unrelated companies. What's the likelihood of all of them making the same mistake on the same generation of cards?

Once upon a time, ALL the people (not just engineers) on earth were thinking that the earth was flat. Only one man countered them. I think the existence of horizon is a strong evidence for this one man who was ignored by everyone else.

Only one man.

 

[AusWolf]

What's the likelihood of AMD simply giving a spec sheet that extends the usable range a little bit for these chips, so AIBs have done what they've been told?

We saw something akin to that with the 5090 and one rare exception having some mitigation for bad power. Again... assumptions don't do you any favors here and they certainly don't support your stance.

I'm not assuming anything. I'm just going by the fact that I haven't heard of a single example of a 9070 card failing because of this.

 

[JustBenching]

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

If every single AMD AIB is wrong with every single one of their cards (which I don't find believable one bit), then we'll get a statement soon, or at least examples of cards failing.

Back in the days, everybody told me that 105 ˚C on my 6750 XT's GPU hotspot was unsafe. The card has been running strong for the last 3 years, with no sign of failure.


The fact that we're talking about dozens of engineers at dozens of different unrelated companies. What's the likelihood of all of them making the same mistake on the same generation of cards?

At this point you are going against physics. Higher temps accelerate electromigration, period. It's not even something you can argue against. Your 6750xt is "fine" for a myriad of reasons. First of all, because gpus and cpus are overvolted by default to account for degradation. So your card might as well have degraded, but it's being fed too much voltage from the factory so you didn't even notice.

 

[AusWolf]

Once upon a time, ALL the people (not just engineers) on earth were thinking that the earth was flat. Only one man countered them.

That's totally unrelated to the current conversation.

 

[Vayra86]

I'm not assuming anything. I'm just going by the fact that I haven't heard of a single example of a 9070 card failing because of this.

Well cue your own argument about how VRAM degradation works then. 'It happens over prolonged periods of time' Which is true.
I'm not saying either that these cards fail out of the box. But I do know and say that high temps kill electronics, and VRAM is a very vulnerable part of those electronics.

That's totally unrelated to the current conversation.

Of course not. Everyone can get it wrong, and peer pressure invites us to.

 

[AusWolf]

At this point you are going against physics. Higher temps accelerate electromigration, period. It's not even something you can argue against. Your 6750xt is "fine" for a myriad of reasons. First of all, because gpus and cpus are overvolted by default to account for degradation. So your card might as well have degraded, but it's being fed too much voltage from the factory so you didn't even notice.

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

 

[Vayra86]

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

... its literally spelled out in his post above. You don't know until you do.
The bottom line remains; you are convinced card is fine, others are not so convinced, you have said card, others do not.

Do the math.

 

[AusWolf]

Well cue your own argument about how VRAM degradation works then. 'It happens over prolonged periods of time' Which is true.
I'm not saying either that these cards fail out of the box. But I do know and say that high temps kill electronics, and VRAM is a very vulnerable part of those electronics.

Then we'll hear about it in due course. Guessing before any news of such occurrences doesn't seem to serve a purpose.

Of course not. Everyone can get it wrong, and peer pressure invites us to.

Yes it is. We're talking about VRAM temperatures, not the Earth being flat or not.

There are also many examples of a single person believing something and actually proven false (look at mental asylums), which is equally unrelated.

... its literally spelled out in his post above. You don't know until you do.

Oh. Sounds very scientific.

 

[Vayra86]

Then we'll hear about it in due course. Guessing before any news of such occurrences doesn't seem to serve a purpose.


Yes it is. We're talking about VRAM temperatures, not the Earth being flat or not.

There are also many examples of a single person believing something and actually proven false (look at mental asylums), which is equally unrelated.


Oh. Sounds very scientific.

Assuming its fine because everyone else said its fine is not scientific either. Its just sheep behaviour and some misguided sense of trust.

I'm not saying these cards are all fubar.

I'm saying we need an official statement on these temps, and until then, all bets are off.
Its really just that. The temp is unusually high and these are not some arcane new version of memory chips or anything. It could do with some reassurance.

 

[AusWolf]

Assuming its fine because everyone else said its fine is not scientific either. Its just sheep behaviour and some misguided sense of trust.

There is no everybody else. I assume it is fine because there is no data to support any claim within this topic, and I'm not an engineer.

I'm saying we need an official statement on these temps, and until then, all bets are off.

Exactly. Until such a statement is made, or we hear about news of cards dying, I have no reason to assume anything.

 

[tugrul_LordOfDrinks]

Until such a statement is made, or we hear about news of cards dying, I have no reason to assume anything.

Space Shuttle Challenger Disaster (1986)

• What happened? The Space Shuttle Challenger exploded 73 seconds after launch, killing all seven astronauts.
• Why? Engineers at Morton Thiokol (the company responsible for the solid rocket boosters) knew the O-rings could fail in cold weather but were pressured by NASA management to approve the launch.
• The consequence? A catastrophic explosion that halted the U.S. space program for years.

Seven astronauts (who died) did not know about those O-rings. They were not informed.

Here is your reason. It's not even "rocket science" to guess this.

 

[AusWolf]

Space Shuttle Challenger Disaster (1986)

• What happened? The Space Shuttle Challenger exploded 73 seconds after launch, killing all seven astronauts.
• Why? Engineers at Morton Thiokol (the company responsible for the solid rocket boosters) knew the O-rings could fail in cold weather but were pressured by NASA management to approve the launch.
• The consequence? A catastrophic explosion that halted the U.S. space program for years.

Here is the reason to not trust. I bet they value memory temperature less than the O-rings of rocket boosters.

Seven astronauts (who died) did not know about those O-rings. They were not informed.

Please bring more examples of completely unrelated engineers failing to design a completely unrelated product and/or pressured by a completely unrelated management team to hit deadlines.

I never said trust AMD, or their dozen or so AIBs. What I said is, there's no data to support any claim here.

 

[Vayra86]

Please bring more examples of completely unrelated engineers failing to design a completely unrelated product and/or pressured by a completely unrelated management team to hit deadlines.

I never said trust AMD, or their dozen or so AIBs. What I said is, there's no data to support any claim here.

And as a result you said to assume they got it right

 

[AusWolf]

And as a result you said to assume they got it right

What else can I assume based on the fact that there isn't a single 9070 with VRAM temps that people here seem to expect, yet not a single one of them has been reported to fail?

I don't want to live my life in fear of something I don't understand. Do you?

Edit: I believe we had this conversation when the GPU hotspot sensor got introduced. Then with 7000-series Ryzen. Nothing blew up. And now, we're having it again.

 

[Vayra86]

What else can I assume based on the fact that there isn't a single 9070 with VRAM temps that people here seem to expect? I don't want to live my life in fear of something I don't understand. Do you?

No, and then we get to the very core of what I'm trying to bring across Since you don't want that, you assume its fine. You assume it because you own the card - I don't assume it because I don't own the card.

Its all psychology. So that is why I've said that statement is so very important. Don't leave this kind of thing up in the air. Nobody benefits.

 

[tugrul_LordOfDrinks]

Please bring more examples of completely unrelated engineers failing to design a completely unrelated product and/or pressured by a completely unrelated management team to hit deadlines.

Astronauts are valuable for space missions. You can't go space without them to do important stuff. Yet they killed them without saying them truth. Now, gaming is not even a major contribution to income for Nvidia, etc. Gaming income for them is expendable. It's not valuable for them. They can go into gpgpu market without gamers. So, its even worse than NASA missions.

They said Titanic was unsinkable... Until it wasn't.

The Boeing 737 MAX Crashes: same thing.

Burnt pci connector: same same. But different.

 

[Vayra86]

Astronauts are valuable for space missions. You can't go space without them to do important stuff. Yet they killed them without saying them truth. Now, gaming is not even a major contribution to income for Nvidia, etc. Gaming income for them is expendable. It's not valuable for them. They can go into gpgpu market without gamers. So, its even worse than NASA missions.

They said Titanic was unsinkable... Until it wasn't.

I'm giving a pattern here. The same pattern found in all catastrophic events.

There is the pattern of poor engineering and people being people.

Another interesting thought line here though is the pattern of 'You won't know until you try'. After all a lot of these failed engineering jobs in history are firsts. Experimentation is extremely human. We learn by failing.

 

[AusWolf]

Astronauts are valuable for space missions. You can't go space without them to do important stuff. Yet they killed them without saying them truth. Now, gaming is not even a major contribution to income for Nvidia, etc. Gaming income for them is expendable. It's not valuable for them. They can go into gpgpu market without gamers. So, its even worse than NASA missions.

They said Titanic was unsinkable... Until it wasn't.

I'm giving a pattern here. The same pattern found in all catastrophic events.

There is no pattern. Just examples of engineers failing and getting things right. There is no evidence to point towards either case regarding the 9070 (if you don't count the fact that NOT ONE card has been reported to fail due to high VRAM temps).