NVidia now HIDING hot spot temperature? A great problem IMO.

[freeagent]

I saw the GN teardown, I am fairly impressed with how it is all pieced together.. pretty clever to get it into 2 slots.

 

[LabRat 891]

I saw the GN teardown, I am fairly impressed with how it is all pieced together.. pretty clever to get it into 2 slots.

Extremely. -and, without too fiddly of disassembly!*.

*funny though... the PCB is so dense, and the thermal interface about as optimal as it can be, I don't see anyone much needing into the assembly (other than waterblock mounting).
Otherwise, it looks 100% built to last the entire period it's designed to. (easy as frig to dust, too).

 

[Dragam1337]

What are you gonna do?

It has LM, It has the fans, the engineering, the materials.

Maybe it was designed that way..

Typical reply from you. The fe uses liquid metal... aibs (gigabyte especially) will 100% use badly applied sh1t tim as always.

And as the tests have shown, the fe cooler is way undersized...who would have thought with a 2 slot cooler for 600 watts...

 

[Neo_Morpheus]

Heh, Nvidia tried many, many times to have a proprietary feature that gamers will buy their GPU's for and hats off to them as they kept at it through thick and thin. They kept failing and consumers never really gave in for a long time. I remember Kyle over at [H] and a few others butchered nvidia so hard in one of the articles they actually took back or never released something dodgy that they were cooking up, can't remember what exactly it was. Don't think it was TWIMTBP but my god was that another shitshow.

I still chuckle at hairworks when i fire up witcher 3, You want nice hair for Geralt? You need an nvidia card for fuck all reason. Of course, just like most of nvidia's proprietary features at the time, it was crap with little to no difference in fidelity (hello RT, hairworks, excessive tesselation etc) for the performance impact. And of course it could be processed with AMD GPU's, just like they locked down bokeh and a few other features to nvidia GPU's only in Just Cause 2. The list is massive.

DLSS and RT are the ones that stuck but nvidia did go all in on both including a name change altogether. DLSS is great as it allowed playable framerates with not much of a visual impact but for me it really depends on the game. I still can't even turn DLSS on in Warzone, FH5 and a couple of titles because it's..shit. But as a whole it was good as it pushed AMD to release FSR which helped a lot of gamers extend the life of their GPU's from both camps.

RT though, don't even get me started. Great for people who like it and more power to them, but for me it's a piece of shit in its present form. It offers very little visual impact for a massive performance drop. Give me better textures instead maybe, those 4k/8k texture mods sure look great in the older games. There's no turning back though and RT is here to stay so how about you do it in moderation and not excessively resulting in 15fps for nvidia and 5 for AMD. Maybe bake it in the engine or something like what UE is doing I suppose. They did the same with tessellation till they realised they were only screwing the consumers and it wasn't helping their bottom line and just went on to the next thing.

Preach it brother!

Your post should be included on all reviews with the hope that we go back to unbiased and real reviews.

Talking about Kyle, funny enough, he turned a blind eye when Ngreedia started including forced telemetry spyware with their drivers without notifying their victims.

 

[freeagent]

Typical reply from you.

I didn't mean to hurt your feelings.

And I was talking about the FE yes.

 

[Dragam1337]

I didn't mean to hurt your feelings.

And I was talking about the FE yes.

Nor did you, but it was just the typical /facepalm moment...

Yeah, which isn't available in most of the world, thus largely irrelevant.

 

[LabRat 891]

The fe uses liquid metal... aibs (gigabyte especially) will 100% use badly applied sh1t tim as always.

I cannot speak for all the AIBs, nor towards any of the prospective RTX5k SKUs, but...
PTM7950 is becoming more common amongst AIBs. The guano AIBs were using, was apparently harming return rates w/in warranty period.

TBF,
Reviews of the partner models will help us tell, but that still doesn't prevent later-production changes...

 

[freeagent]

Nor did you, but it was just the typical /facepalm moment...

Yeah, which isn't available in most of the world, thus largely irrelevant.

Not sure about you, but I have no intention of buying one, but the tech that went into building one is pretty cool.

 

[AsRock]

Just maybe the GPU temp is the hotspot temp, and ditched the other.

EDIT

I saw the GN teardown, I am fairly impressed with how it is all pieced together.. pretty clever to get it into 2 slots.

Yes very impressive but at the same time a nightmare, liquid metal rubber seal and ribbons.

 

[Dragam1337]

Not sure about you, but I have no intention of buying one, but the tech that went into building one is pretty cool.

Ofc you don't, as you never buy halo cards, which is fine, but it also makes your opinion on the matter... less interesting.

Knowing that tim is correctly applied will matter to people running a 600 watt chip, and hot spot sensors greatly help in that regard.

As for fe being interesting... not really. Imo it's by far the least interesting 5090. Even if it was available outside 'murica, i would never buy one. Aib versions are just straight up better... as always.

 

[freeagent]

Ofc you don't, as you never buy halo cards, which is fine, but it also makes your opinion on the matter... less interesting.

Dude I have had so many halo cards. Think halo cards are a new thing? This has been my hobby for over 20 years.

Knowing that tim is correctly applied will matter to people running a 600 watt chip, and hot spot sensors greatly help in that regard.

Watch the metrics, easiest way to tell. You don't have to open it up. I haven't opened my 3070Ti, still runs mint.

As for fe being interesting... not really.

Well.. that is just your opinion, we all have one

 

[RJARRRPCGP]

I rather have an AIB with thin TIM, than an FE with a soldered-on faulty heat sink chamber, FFS!

 

[Dragam1337]

Dude I have had so many halo cards. Think halo cards are a new thing? This has been my hobby for over 20 years.

Watch the metrics, easiest way to tell. You don't have to open it up. I haven't opened my 3070Ti, still runs mint.

Well.. that is just your opinion, we all have one

Sure.

Yes, watching the metrics is the easiest way to tell - the hot spot temp being the primary one, which you are saying is fine that it's being removed... i swear dude, every time you reply...

 

[freeagent]

Sure.

Yes, watching the metrics is the easiest way to tell - the hot spot temp being the primary one, which you are saying is fine that it's being removed... i swear dude, every time you reply...

That is indeed my opinion, as someone who is actually gonna buy a 5090... aka worth a bit more than someone who aint actually gonna buy one.

Well, it is clear you know what you are talking about. Enjoy.

 

[evernessince]

My only concern is that engineered life may be well-shorter than we've long come to expect.

NGL, seeing the 5090 FE teardown and performance #s has me intrigued.
$900-1100 for an XTX or $2k*+ for a 5090 that performs ~67-100% faster than the 7900 XTX? (*Pipedream, ofc; scalpers gon a scalp)

The 12V2X6 connector, the cables connecting the PCBs in the FE, hiding of hotspot temps, and the liquid metal would all have me concerned for longevity.

As Derbauer has pointed out in past videos you can either have a quality nickle plating at the cost of thermal performance or a thinner nickle plating that's much more likely to be eaten through by liquid metal. I can't say I trust all AIBs or even Nvidia to make the right choices there given that's a longevity issue likely to be a problem outside of warranty unless some really bad choices were made.

Doing PCIe 5.0 over cable is extremely hard and cables in general have a higher failure rate than data going directly through the PCB.

Cards with uneven TIM / LM application will slip through QA as always and with hidden hotspot temps that just means that a slightly greater proportion will die prematurely as a result.

Couple that with the elevated failure rate of the 12V2X6 connector and everything else above and it appears to me that sacrifices are being made in terms of long term service life. If I had to purchase a 5090, I'd grab an AIB model without LM and power limit it.

 

[BoggledBeagle]

Yes very impressive but at the same time a nightmare, liquid metal rubber seal and ribbons.

I watched GN teardown and it seems that the liquid metal breached one or two ribbons, they probably overfilled it, or they have problem with air pressure in the chip area while mounting the cooler.

How I wrote in the thread about the FE cooler, their cooler has about half surface area of the larger AIB models, and the liquid metal is a desperate measure to lower the temps the chip is hitting with such a small surface area cooler. And the liquid metal was not enough, they simply need to hide the evidence of what is going on.

I do not think that making the 600W card so small was a good decision, they insisted on carrying this plan out even though they found out that the card overheated, and the disastruous fallout for everybody is that they decided to hide hot spot temperature.

 

[AleXXX666]

We did not have hotspot sensors until the 30 series and it was just fine? I mean 1080Tis / 1060s etc are still running just fine

unless they were used in mining...

 

[BoggledBeagle]

.... For example my current 4070 started with the hot spot being about 10°C above the whole chip temperature, but then it slowly started drifting higher and now is 30°C above. My card is probably affected by the crap-paste problem Igor reported about, and now thanks to the hot spot temperature I know that I should probably fix it.

I checked the card again and now it is already hitting 103°C... And that is a 4070 with a size of 4090. Hot spot temperature IS CRITICAL to be able to tell that there is something wrong going on with your card.

 

[AusWolf]

Users on both sides are fanatics, hellooo do you see where we are?

Is this a place where normal people come to hang out?

No? That's because we are all a bunch of nerds or geeks, or nerds with geek-like tendencies.

Being a nerd or geek doesn't mean you have to be fanatic about a brand. Quite the opposite, in my opinion. It means being fanatic about tech of all brands and colours. Right?

Disclaimer: It's not the same thing as having preferences. Of course everybody has preferences. But being fanatic? Why?

 

[agent_x007]

Worst case option : VRAM temp is new "GPU edge" temp.

 

[freeagent]

Being a nerd or geek doesn't mean you have to be fanatic about a brand. Quite the opposite, in my opinion. It means being fanatic about tech of all brands and colours. Right?

Disclaimer: It's not the same thing as having preferences. Of course everybody has preferences. But being fanatic? Why?

It was a joke.

 

[AusWolf]

It was a joke.

Fair enough. Faith in humanity is restored.

 

[mclaren85]

I would ignore all temperature readings if I were you. Some might remember that I was so obsessed with the x570 chipset temperature and believe me the chip would last more than it become obsolete.

 

[freeagent]

That guys hotspot does suck though.

 

[JustBenching]

Another ngreedia planned obsolescence thread. Gotta love them

Engineers know everything, they never make mistakes, and certainly his opinion as an employee is above everyone else, safety > profit. Stop complaining. After all, there is no precedent for planes crashing, cars exploding, smartphone batteries catching fire and CPUs degrading. None of this was caused by human error, I promise.

But none of those issues were fixed by a guy posting on a forum..

An Appeal to Authority fallacy is no substitute for an argument. As Denver pointed out, being a specialist in a field does not make you above reproach

Appeal to authority is not a fallacy, I've seen a lot of people confused about this. The fallacy is the appeal to a false authority. Like, the 5090 is fine cause Messi said it's fine. Messi isn't an authority on GPUs, so the argument is fallacious. Appealing to nvidia on matters of GPU is not fallacious.