CheeseballI do not consider GPU Tweak III as bloatware as all the functions in the software are only for monitoring or tweaking ASUS graphics cards. I would consider Armory Crate as bloatware.

GPU Tweak III may actually be better than MSI Afterburner/RivaTuner Statistics Server in this regard.

I also consider Afterburner bloatware. All that functionality should be just basic driver thing. It's just my opinion, though.

AusWolfI also consider Afterburner bloatware. All that functionality should be just basic driver thing. It's just my opinion, though.

Tweaking (e.g. overclocking, memory tuning, etc.) should NOT be part of basic driver control panel functionality unless its blocked off with legal warnings (like how it's done in Adrenalin). Monitoring sensors should be fine since those are usually built into the GPU and shouldn't run any risk.

CheeseballTweaking (e.g. overclocking, memory tuning, etc.) should NOT be part of basic driver control panel functionality unless its blocked off with legal warnings (like how it's done in Adrenalin). Monitoring sensors should be fine since those are usually built into the GPU and shouldn't run any risk.

Tweaking should be part of the basic driver package, with legal warnings of course - that's my whole point. It's your GPU, your choice how to use it.

KwadratowiczAnd one more thing to think about, there is info that the plastic that is used in the connector on the side of the board has higher melting temperature than connector from the psu or adapter (Der8auer was talking about it, and from looking at the connectors it's clear what side is weaker to temp), that's why it's always adapter or psu cable that is accused of being bad, and with that They shift the blame from bad standard to cables.

Logically if adaptor was to blame we would see burnout on the adaptor connection point, but its always on the slot area on the GPU, so I dont see how a new gen of PSU removes the risk.

Waldorf@noel_fs
because the part comes with the card?
the same way you mention the car brand, if they deliver the car with tires that fail prematurely.

what?

in that case you would mention the brand of the tires in the title, and the specifics of the car in the article.

im smelling some shady stuff, if i understand correctly the card is in perfectly working condition if there is anything failing its not the cards fault


anyway im done with this, in the last year ive seen way too many questionable articles in the site so ill just move on from it

I mean I guess it was only a matter of time. I specifically went with the Sapphire Pure OC 9070XT, the feature list is all but identical minus 2 power phases and some bling, but 2x8 pin. It boosts easily to almost 4Ghz with just an UV and PL increase and has to be the coolest running card with such great performance I have ever owned. I was a very hardcore Nitro user had a 5xxx and a 6xxx made a jump to Powecolor for 7xxx but that connector despite my PSU having one was just a NOPE...

AusWolfNo, I believe my eyes. I never had any issue with any driver for RDNA 2, and I don't have a reason to believe that it'll change in the future.

You still haven't answered my question. Why talk about this here? There are more appropriate places to vent your unhinged hatred towards anything bearing the colour red, you know.

RDNA2 drivers has been handled mostly fine and they have made various improvemnts with Adrenaline across the years, but I think it's a bit disigenuous to take as if all has been fine. As a very good example OpenGL performance at launch and in general for Radeon products was awful for various years until they fixed it, the performance was so bad that on lower and mid end RX 6000 RDNA GPUs I could observe an Intel UHD 630 performing way better in OpenGL until it was fixed.

I have also had minor problems at times even if rarely with some applications (not games) there and there, for which I actually don't have any complaints as I reported to AMD and they usually fixed them in the next release, but I do wonder if somethinhg similar would arise if now they would still fix so in time or would leave to maybe some day, some month, because of the new scheme. I have some disbeliefs with AMD on this as their statement has been very similar to what they said for Polaris and Vega some years ago and they haven't handled them that well, but time will tell.

Hopefully they do use this as they said for general positive effects in newer and future products and do keep the RDNA1 and RDNA2 branch properly maintained instead of how they have handled Vega and Polaris one kind of forgotten to the side and several times between very long period of times for some more important problems. But only time will be able to show and tell from AMD's output on this and future products and I will judge them at that point based on so, but will keep with some understandably caution based on the previous output they have had.

this card consumes only half for what it's rated for (300W) and yet still manages to burn pins. It's a fantastic product released or pushed by Nvidia which has seen the most amount of PC fires in years now. It's time Nvidia is held responsible and starts reverting back since the need or use of special devices to "protect" the wires from getting too hit. It's stupid.

Even a PCI-E 8 pin can handle 25x insertion and release without even corroding. And still its more robust then this solution is.

TheinsanegamerNGPUs also pull power from motherboards and use 12v for signaling.

PCIe doesn't use 12V for signaling. And while GPUs can pull power from the motherboard, my guess for a lot of higher end cards, slot power is just to get the controller chip on the card up and running, which that turns everything else on.

EDIT: Out of curiosity, I fired up a HWiNFO since it can read the power rails on my video card and put a reasonably good load on it. Out of the 235W it was consuming, ~210W was coming from the 12VHPwr cable. So the slot isn't really providing much power. And I'd imagine this behavior is similar to all other higher end models.

TheinsanegamerNIt's not that easy. You need to upgrade the backend bus architecture to the same voltage. Or you need to iclude an expensive VRM system to convert voltage back and forth, which takes up a lot of space and makes a lot of heat.

Any higher voltage handling would only be needed on the supplemental power side. Considering this would be a sink-only system, there's no need to convert voltages back and forth. Plus GPUs don't directly operate at 12V anyway, otherwise there wouldn't be a need to have the VRM. At least one with buck converters.

Speaking of which, PCIe SIG standardized a 48V connector, which seemed to fly under the radar.

EDIT the second: Well since finding a GPU datasheet is damn near impossible to find, why not one for GDDR6? As far as I can tell, the V_DD on the thing is 1.35V.

RulumiRDNA2 drivers has been handled mostly fine and they have made various improvemnts with Adrenaline across the years, but I think it's a bit disigenuous to take as if all has been fine. As a very good example OpenGL performance at launch and in general for Radeon products was awful for various years until they fixed it, the performance was so bad that on lower and mid end RX 6000 RDNA GPUs I could observe an Intel UHD 630 performing way better in OpenGL until it was fixed.

I have also had minor problems at times even if rarely with some applications (not games) there and there, for which I actually don't have any complaints as I reported to AMD and they usually fixed them in the next release, but I do wonder if somethinhg similar would arise if now they would still fix so in time or would leave to maybe some day, some month, because of the new scheme. I have some disbeliefs with AMD on this as their statement has been very similar to what they said for Polaris and Vega some years ago and they haven't handled them that well, but time will tell.

Hopefully they do use this as they said for general positive effects in newer and future products and do keep the RDNA1 and RDNA2 branch properly maintained instead of how they have handled Vega and Polaris one kind of forgotten to the side and several times between very long period of times for some more important problems. But only time will be able to show and tell from AMD's output on this and future products and I will judge them at that point based on so, but will keep with some understandably caution based on the previous output they have had.

You just said it yourself that the issues have been fixed. That's why they're moving those drivers to a different channel.

@noel_fs
Did the user buy the adapter, or a gpu that had the adapter coming with it? right.
Its important for (possible) buyers to know WHICH card/brand can have problems, so they avoid it (or not), and how would that work, without naming brand/model? right.

but yeah, stay away from the bad articles 😂

shuffy_losysPCIe doesn't use 12V for signaling. And while GPUs can pull power from the motherboard, my guess for a lot of higher end cards, slot power is just to get the controller chip on the card up and running, which that turns everything else on.

EDIT: Out of curiosity, I fired up a HWiNFO since it can read the power rails on my video card and put a reasonably good load on it. Out of the 235W it was consuming, ~210W was coming from the 12VHPwr cable. So the slot isn't really providing much power. And I'd imagine this behavior is similar to all other higher end models.


Any higher voltage handling would only be needed on the supplemental power side. Considering this would be a sink-only system, there's no need to convert voltages back and forth. Plus GPUs don't directly operate at 12V anyway, otherwise there wouldn't be a need to have the VRM. At least one with buck converters.

Speaking of which, PCIe SIG standardized a 48V connector, which seemed to fly under the radar.

EDIT the second: Well since finding a GPU datasheet is damn near impossible to find, why not one for GDDR6? As far as I can tell, the V_DD on the thing is 1.35V.

Looks more sturdy. It's also part of 5.1 spec so who knows, maybe 60-series will opt for this.

AusWolfI have one problem with this.

Why not improve efficiency instead of churning out 450-600 W cards?

"Oh look, our new 9989 XT Ultra Pro Ti is so fast!" - Yeah, right! :rolleyes:

If hardware makers stayed in the land of sanity with power consumption on their cards, we would have never needed a new connector to replace 3x old ones in the first place.
2x 8-pins should be more than enough for any GPU.

Why not stop capitalism? That's what you asked here. I don't know why, fact is, its reality.

If everybody loved each other, we wouldn't need guns is another one of those beautiful utopian oneliners.

AusWolfMore vendor-specific bloatware for something that should just work? Oh god no! :fear:

Yeah I've never quite grasped how it works in peoples' heads when they say 'get an Astral' so you can see how poorly the design actually performs. The GPU still hasn't got jack shit to help it in longevity. You can just pre empt the eventual disaster better.

Vayra86Why not stop capitalism? That's what you asked here. I don't know why, fact is, its reality.

If everybody loved each other, we wouldn't need guns is another one of those beautiful utopian oneliners.

I could have asked that, but didn't.

Since "global warming" is such a serious issue, then maybe hardware manufacturers could put their act together and focus on efficiency for a change. Sure, it's utopia, but just saying...

AusWolfI could have asked that, but didn't.

Since "global warming" is such a serious issue, then maybe hardware manufacturers could put their act together and focus on efficiency for a change. Sure, it's utopia, but just saying...

Hey I don't disagree, but I spot the total opposite movement: nationalism, grabbing guns, and killing the other off instead of coming together for solutions. And ironically most of that killing is somehow related to the West.

Vayra86Hey I don't disagree, but I spot the total opposite movement: nationalism, grabbing guns, and killing the other off instead of coming together for solutions. And ironically most of that killing is somehow related to the West.

Unfortunately, I don't disagree, either. Only the common man gets the "global warming" and "do your part for the planet" talk. :(

JismIt's a fantastic product released or pushed by Nvidia which has seen the most amount of PC fires in years now.

Most fire hazards originate from power supplies of unknown origin.

JismIt's stupid.

My opinion? The ‘handful’ of customers affected by the problem are the typical Error 40 cases, and instead of admitting this, they are trying to blame NVIDIA. It's simple; case closed!

JismEven a PCI-E 8 pin can handle 25x insertion and release without even corroding.

And this too will fail prematurely when it comes to cable management ‘bending and breaking’.

@EsaT

With a friendly smile,
Fred.

Hecate91Strawman arguments, how do those work?

The point is the connector shouldn't be failing in the way it is, and this "but the sales numbers bro" nonsense is probably a reason why no one cares enough to make a better power connector when marketing and sales managers tell the engineers no, and why make it better when you can just hide the real issue with telling people they're plugging it wrong and they should buy a new PSU every time they buy a new GPU.

The point is, the connector is failing the same way the 8 pin connector failed, and anyone crying about it being a failed esign has nothing but speculation to sit on.

Hecate91Have any hard evidence on how many 8 pin connectors fail then? Or do you just want to a defend mega corporation for a shitty design?

You're the one claiming the 12v connector fails more often. Provide evidence.

Hecate91This makes no sense at all, and the 8 pin connector is so overbuilt it wouldn't melt because the pins are thicker, it doesn't take an EE degree to figure out thicker metal pins and wires equals more durable.
And no that's not how a safety margin works, having a safety margin means the connector is built better than it needs to be in an amount over the rated specs, not just functioning to the bare minimum of the spec. It's like expecting seatbelts alone to protect you when blind spot warning, auto braking, seatbelts and airbags are significantly better protection.
The wikipedia article points out how the safety margin on 12vhpwr is much lower. 1.68 for 8 pin molex, 1.14 for 12v2x6.
en.wikipedia.org/wiki/12VHPWR

This shows blatant ignorance to standards.

The 8 pin was rated at 330w maximum. If GPU makers chose to run it at 300w instead of 150w, it would have similar safety margins to 12v, which you guys cant stop crying about.

Hecate91The connector was sponsored, in other words paid to be approved and pushed to market by Nvidia and Dell.
I'm reposting this article that @evernessince posted above, since people still want to ignore it. PCI-SIG admits themselves that members are responsible for testing the standard themselves, it absolutely baffles me how a standards organization responsible for these standards doesn't bother to verify the sponsored connector for safety and quality vs. existing standards.
arstechnica.com/gadgets/2022/12/pci-standards-group-deflects-assigns-blame-for-melting-gpu-power-connectors/

Ah, there's the crux of the issue, 12v bad because nvidia bad.

Hecate91AMD isn't forcing their partners to use 12v2x6, but for some reason Sapphire thought it would make their card look better, or assumed it would be fine at 350W but of course it's not.

It's fine for a 350w card. Sapphire failed to put load balancing on their GPU and paid the price.

Hecate91You mean in a failure state that shouldn't even be happening if the protections to prevent it from becoming unbalanced were apart of the standard, not needing to buy a $100 monitoring device to know when the connector is about to melt.

Again, has has been said for the 1000th time, it is not on the connector to do load balancing, it is on the device pulling power. That is how the whole industry, not just DIY PCs, operate.

Speaking of which, where are all the AI servers burning out their hundreds of thousands of GPUs that they are hammering 24x7, al with 12v power connectors? We should be seeing warehouses full of failed cards....where are they?

Hecate918 pin connectors didn't fail in the same way unless using a dumpster fire of a PSU, and yet reports of 12 pin connectors is happening almost every day, despite we're told the connector is better than the old 8 pin connector because it's smaller and more convenient to plug in than 2 or 3 8 pin connectors.

More qualifying statements. Not a single hard number.

What that tells me is you dont have any actual numbers, no evidence, just feelings. And because nvidia bad because green bad, and therefore anything they make is bad, you label 12v as bad and use that to color your opinion irregardless of evidence, or the lack thereof.

TheinsanegamerNThe point is, the connector is failing the same way the 8 pin connector failed, and anyone crying about it being a failed esign has nothing but speculation to sit on.

No, the point is the connector is failing more often than the 8 pin one did, except the 8 pin molex has been the standard for much longer.

TheinsanegamerNYou're the one claiming the 12v connector fails more often. Provide evidence.

You claimed the 8 pin connector fails too, but I have yet to see any evidence the 8 pin connector has just as many failures.
There have been plenty of reports on reddit, various news articles on 12 pin connectors melting, tech Youtubers have covered it quite a bit as well, the evidence is there that the connector just sucks and yet people still have some weird defense for a badly designed cable.

TheinsanegamerNThis shows blatant ignorance to standards.

The 8 pin was rated at 330w maximum. If GPU makers chose to run it at 300w instead of 150w, it would have similar safety margins to 12v, which you guys cant stop crying about.

The 8 pin can handle 330W, that is double the power of what it's rated for, meanwhile 12vhpwr/ 12v2x6 doesn't have any significant safety margin above the 600W rating.

TheinsanegamerNAh, there's the crux of the issue, 12v bad because nvidia bad.

It's bad and the company who designed it should take some responsibility for it, that's the issue.
Not the childish "boo hoo people are blaming Nvidia".

TheinsanegamerNIt's fine for a 350w card. Sapphire failed to put load balancing on their GPU and paid the price.

Load balancing isn't a part of the specification of 12v2x6, even the Asus ROG Astral with per pin power detection is likely violating the PCI-SIG specs by going beyond what Nvidia designed.

TheinsanegamerNAgain, has has been said for the 1000th time, it is not on the connector to do load balancing, it is on the device pulling power. That is how the whole industry, not just DIY PCs, operate.

Speaking of which, where are all the AI servers burning out their hundreds of thousands of GPUs that they are hammering 24x7, al with 12v power connectors? We should be seeing warehouses full of failed cards....where are they?

And for the 10000000000th time, the connector wasn't designed with load balancing, when the connector is a fire risk it should be on the connector and the PSU to make sure the GPU side of the connector doesn't turn into a mess of melted plastic.
Server hardware isn't the same as consumer parts, this much should be obvious, Nvidia took the cheaper route with the consumer cards and it shows when comparing the connector to 8 pin molex.

TheinsanegamerNMore qualifying statements. Not a single hard number.

What that tells me is you dont have any actual numbers, no evidence, just feelings. And because nvidia bad because green bad, and therefore anything they make is bad, you label 12v as bad and use that to color your opinion irregardless of evidence, or the lack thereof.

So any hard numbers on 8 pin molex failing then?
Again there aren't any numbers on this because any Nvidia partner making GPUs or PSUs would never admit what the failures really are, and the victim blaming is just weird, you know defending a company for this doesn't benefit you, and the blanket defense for a badly designed connector and a multi-trillion dollar company doesn't help either.

TheinsanegamerNWe should be seeing warehouses full of failed cards....where are they?

The 12VHPWR connector will work under normal conditions!

Power supply manufacturer Seasonic still recommends not bending the cables at all, and especially not directly in the area of the connector.

Source ComputerBase (de)

Just today, I saw the opposite again in the bubble.😒 And if ‘normal conditions’ are not adhered to here, one can expect far less from Reddit DAU users, right? Right?

With a friendly smile,
Fred.

FR3D1The 12VHPWR connector will work under normal conditions!

Which didn't affect the 8 pin molex connector. If not being able to bend the connector at all is the "normal condition" then the standard needs to be revised so the connector can be used in a horizontally installed GPU in an average sized ATX case.

FR3D1The 12VHPWR connector will work under normal conditions!

Source ComputerBase (de)

Just today, I saw the opposite again in the bubble.😒 And if ‘normal conditions’ are not adhered to here, one can expect far less from Reddit DAU users, right? Right?

With a friendly smile,
Fred.

Except 'normal conditions' is a word that had incurred some limitations since 12VHPWR- and therein lies the problem.

ChomiqLooks more sturdy. It's also part of 5.1 spec so who knows, maybe 60-series will opt for this.

The problem is no PSU maker signed up for it on the ATX side of things. Or has plans to AFAIK.

My guess is if it's being used or will be used, it's in compute farms.

The 12 pin clearly fails at a higher rate than the 8 pin, validated by extensive media coverage from Gamers Nexus, Debeaur and others, including multiple failure analysis labs reports. You are being deliberately obtuse and clearly trolling by asking for evidence repeatedly and aggressively. Educate yourself and stop spreading misinformation. 12 pin in all forms is dangerous and insufficient as an electrical standard. The standards organization is controlled by industry and as Nvidia led development of the standard this is their fault to own.

GenericNinjaThe 12 pin clearly fails at a higher rate than the 8 pin, validated by extensive media coverage from Gamers Nexus, Debeaur and others, including multiple failure analysis labs reports. You are being deliberately obtuse and clearly trolling by asking for evidence repeatedly and aggressively. Educate yourself and stop spreading misinformation. 12 pin in all forms is dangerous and insufficient as an electrical standard. The standards organization is controlled by industry and as Nvidia led development of the standard this is their fault to own.

Wasn't 12pin originally developed for Intel spec?