very nice and long overdue.

How long until we see one of these go up in flames?

Cottage industry of adding safety features to this fire hazard is thriving.

Pointless unless each of the six 12V pins has it's own sensor right on the pin, which is impossible to begin with.

And these marketing bullshit products prove we need to stick with 8-Pin, which was never an issue to begin with. Let the business server market do all they want, but hardware for the average joe should be safe.

rackster37How long until we see one of these go up in flames?

You know, after all of the armchair engineering, debating, snickering, the works. The connector itself will always fail because it is the singular point of failure, and at high wattages these failures manifest themselves pretty violently. Who'd have thought 600 W through a single cable would end up brutally exposing poorly built power supplies with little to nothing in the way of safety measures - and for many reasons, the connector takes the fall.

@Dr. Dro
That, and there is a simple fact that the connector just works with the hand it has been dealt - which is itself shit. The PCI-E AIB form-factor has NEVER been intended for 600W guzzling boards. It's an issue which basically is impossible to solve at this point - if we want to continue with ever increasing power budgets, then a new standard for GPUs specifically would have to be drafted. This is unsustainable.

Dr. DroYou know, after all of the armchair engineering, debating, snickering, the works. The connector itself will always fail because it is the singular point of failure, and at high wattages these failures manifest themselves pretty violently. Who'd have thought 600 W through a single cable would end up brutally exposing poorly built power supplies with little to nothing in the way of safety measures - and for many reasons, the connector takes the fall.

It's been proven time and time again that the connectors are the issue... but yeah bro it's definitely the "poorly built power supplies."

I don't understand why you feel the need to glaze a multi-billion dollar company that has clearly made a mistake and refuses to admit it.

Just put IEC C22 sockets on the videocards already.

Mmm, 90° connectors do help with the bend issue and older cases that don't have space between GPUs and side panels and look better. However, I think maybe 80-85° may better for the rare scenario where there actually is another AIB below the GPU.
At least ASRock is helping the situation somewhat, but I think ASUS's BTF in my opinion is the better solution. A 12v-2x6 connector doesn't have to be the pass through on the back or front of the motherboard, it could be 8pins/6pin connectors.

rackster37It's been proven time and time again that the connectors are the issue... but yeah bro it's definitely the "poorly built power supplies."

I don't understand why you feel the need to glaze a multi-billion dollar company that has clearly made a mistake and refuses to admit it.

The only thing that has been proven time and time again is that 2x6 has greatly improved the contacts, and that this isn't an Nvidia issue (and I'm tired of people pretending it is). The first few documented cases of Radeons that blew up with the connector are already in our hands. For example, the guy whose Taichi 9070 XT caught fire was running an underdimensioned 700 W power supply.

It doesn't take much of a genius to figure it out, the moment anything goes wrong, current follows the path of least resistance.

Dr. DroThe only thing that has been proven time and time again is that 2x6 has greatly improved the contacts, and that this isn't an Nvidia issue (and I'm tired of people pretending it is). The first few documented cases of Radeons that blew up with the connector are already in our hands. For example, the guy whose Taichi 9070 XT caught fire was running an underdimensioned 700 W power supply.

It doesn't take much of a genius to figure it out, the moment anything goes wrong, current follows the path of least resistance.

Okay, so the connector sucks on both Nvidia and AMD cards, what's your point?

I'll shut up if you give me a valid source that says power supplies really are the issue, not just "trust me bro."

Onasi@Dr. Dro
That, and there is a simple fact that the connector just works with the hand it has been dealt - which is itself shit. The PCI-E AIB form-factor has NEVER been intended for 600W guzzling boards. It's an issue which basically is impossible to solve at this point - if we want to continue with ever increasing power budgets, then a new standard for GPUs specifically would have to be drafted. This is unsustainable.

Yea, the connector can't load balance the 12V leads in the cables. It's a design flaw as it stands currently.
Either have load balancing on the card on each lead or/and have proper voltage monitoring via the sense pins to the PSU, these are the most prominent current solutions.

600+ Wattage by itself doesn't matter that much. Connectors also melted on 4070 Ti SUPERs (300W), 4080s (350W), 4080 SUPERs (375W), even on a 9070XT (350W) just recently.
As long as a fault leads to a single wire running the whole load, stuff will get hot. :(

ASRock reminds users to follow their GPU maker's power recommendations, to fully seat connectors until they click, and to avoid bending or stressing the cable near the plug.

Far too complicated.

Let me say it again: a coloured connector is neither a safety measure nor a solution to the problem.

--

Nice marketing. People will buy.

--

Where is the temperature limit set?
How frequently is it polled?
Is the power cut from the graphic card when triggered?
What happens when the power is cut - do we get a blue screen of death or an X-server lockup?

What is the recommended course of action when this temperature feature triggers multiple times a day?


How is the user notified?
How will the user know the blue screen of death was caused by the temperature sensor in the cable?


So many questions and so few answers in this thread.

rackster37Okay, so the connector sucks on both Nvidia and AMD cards, what's your point?

I'll shut up if you give me a valid source that says power supplies really are the issue, not just "trust me bro."

The man who runs this channel, Aris, is one of the leading experts in power supplies. Former reviewer at TPU and responsible for Cybenetics certification system. I recommend starting with this video, but watch all of the videos on 2x6 that he has on his channel and website. He's fully debunked and explained the whole thing at large, and these 2x6 connectors are safe and that's a hill I'm willing to die on.


The problem goes far beyond just this connector: draws are way to too high for the form factor and it remains this is a singular point of failure situation.

Onasi@Dr. Dro
That, and there is a simple fact that the connector just works with the hand it has been dealt - which is itself shit. The PCI-E AIB form-factor has NEVER been intended for 600W guzzling boards. It's an issue which basically is impossible to solve at this point - if we want to continue with ever increasing power budgets, then a new standard for GPUs specifically would have to be drafted. This is unsustainable.

Agree, personally I like ASUS' BTF, I'd actually be on board if they opened the standard and shared it with the rest of the industry. It's solid and remained cool even with 1000W+ draws.

Dr. DroThe only thing that has been proven time and time again is that 2x6 has greatly improved the contacts, and that this isn't an Nvidia issue (and I'm tired of people pretending it is). The first few documented cases of Radeons that blew up with the connector are already in our hands. For example, the guy whose Taichi 9070 XT caught fire was running an underdimensioned 700 W power supply.

A power supply unit with a given Wattage is the root cause?
Why should a wattage rating of a power supply cause the issue?

I assume we talk about this gpu:www.asrock.com/Graphics-Card/AMD/Radeon%20RX%209070%20XT%20Taichi%2016GB%20OC/#Specification

850 Watt recommendation. I read quite often that the graphic card makes power supply recommendation is always far bigger as needed when "using" a "quality" power supply unit.

Are you implying the power supply in question delivered too much Voltage and current to that graphic card with the underrated 700 Watts rating?

Help me understand your point, please.

_roman_A power supply unit with a given Wattage is the root cause?
Why should a wattage rating of a power supply cause the issue?

I assume we talk about this gpu:www.asrock.com/Graphics-Card/AMD/Radeon RX 9070 XT Taichi 16GB OC/#Specification

850 Watt recommendation. I read quite often that the graphic card makes power supply recommendation is always far bigger as needed when "using" a "quality" power supply unit.

Are you implying the power supply in question delivered too much Voltage and current to that graphic card with the underrated 700 Watts rating?

Help me understand your point, please.

That seems to be the case, underdimensioned and potentially, low quality model. Don't remember exactly which one it was.

But given the nature for this cable, I'd only be comfortable with ample headroom and perfect mechanical connection on both ends.

At the point I've lost count of the number of times they've supposedly fixed the issue.

Adding a resistor isn't a fix to a flawed connector design. But interesting to see third parties add things like temperature protection which should have been an included feature.
Is the resistor reporting temps back to the PSU?
At what temp does it trigger a shutdown on the 12v cable?
BTW, yes this is on Nvdia, as Nvidia and Dell Technologies are the main contributors to the 12hpwr specification.

Dr. DroWho'd have thought 600 W through a single cable would end up brutally exposing poorly built power supplies with little to nothing in the way of safety measures - and for many reasons, the connector takes the fall.

PSU has nothing to do with this.
Only way for PSU to protect this badly designed 12VHPWR/12V-2×6 connector would be by limiting whole 12V output to something like little over 10A.

Dr. DroFor example, the guy whose Taichi 9070 XT caught fire was running an underdimensioned 700 W power supply.

PSU doesn't push any current, it will only provide/supply current when something in other end of wire draws/pulls that current.

And if PSU is undersized and load draws too much current/power for the PSU, in that case PSU's protections activate and it shuts down.
Or if voltage becomes too unstable, then it's pretty much guaranteed crashing of GPU/something else...
So basically that too weak PSU would be safer for this connector than beefy high quality PSU capable to actually providing some amount more over numbers in sticker.

PSU doesn't care (and can't even know) where the current goes for as long as it stays inside its total maximum output.
Just like your power company doesn't know and couldn't care less about how your home consumes that power, being only interested that you don't try to melt their power line to your house/stay inside limits of connection you're paying for... And if your own in house electric appliances burn your house, that's in no way their fault.


And even if PSU is bad and lacks working overcurrent/power protections, laws of the physics still apply and it can't push more current out than that drawn by connected device.
Unless it's the cable of PSU which is shorting, which hasn't been the case in any of these melting connectors.

Dr. Dro...these 2x6 connectors are safe and that's a hill I'm willing to die on.

It would be safe if used inside its actual capabilities, which are significantly weaker per pin than for old 8pin PCIe power connector instead of higher!
But Nvidia Stockton Rushed their own MIASS specs for it, basically removing all proper safety margins and then not even mandating any current balancing.

EsaTPSU has nothing to do with this.
Only way for PSU to protect this badly designed 12VHPWR/12V-2×6 connector would be by limiting whole 12V output to something like little over 10A.

PSU doesn't push any current, it will only provide/supply current when something in other end of wire draws/pulls that current.

And if PSU is undersized and load draws too much current/power for the PSU, in that case PSU's protections activate and it shuts down.
Or if voltage becomes too unstable, then it's pretty much guaranteed crashing of GPU/something else...
So basically that too weak PSU would be safer for this connector than beefy high quality PSU capable to actually providing some amount more over numbers in sticker.

PSU doesn't care (and can't even know) where the current goes for as long as it stays inside its total maximum output.
Just like your power company doesn't know and couldn't care less about how your home consumes that power, being only interested that you don't try to melt their power line to your house/stay inside limits of connection you're paying for... And if your own in house electric appliances burn your house, that's in no way their fault.


And even if PSU is bad and lacks working overcurrent/power protections, laws of the physics still apply and it can't push more current out than that drawn by connected device.
Unless it's the cable of PSU which is shorting, which hasn't been the case in any of these melting connectors.


It would be safe if used inside its actual capabilities, which are significantly weaker per pin than for old 8pin PCIe power connector instead of higher!
But Nvidia Stockton Rushed their own MIASS specs for it, basically removing all proper safety margins and then not even mandating any current balancing.

I'm not even going to address the conspiracy theories and unfounded accusations, I've better things to do with my limited time and internet access right now but this is just a compendium of the anti-Nvidia, armchair engineering rhetoric spewed on Reddit as gospel. I'm just going to ask:

How on Earth does a power supply have nothing to do with how power is delivered and distributed? :confused:

Dr. DroThe man who runs this channel, Aris, is one of the leading experts in power supplies. Former reviewer at TPU and responsible for Cybenetics certification system. I recommend starting with this video, but watch all of the videos on 2x6 that he has on his channel and website. He's fully debunked and explained the whole thing at large, and these 2x6 connectors are safe and that's a hill I'm willing to die on.

The tests with brand new connectors are useless. Try testing with old oxidised connector or socket with resistances on some pins even ten times higher than new. Problem arise with time when some pins have very different resistances than others.

Dr. DroI'm not even going to address the conspiracy theories and unfounded accusations, I've better things to do with my limited time and internet access right now but this is just a compendium of the anti-Nvidia, armchair engineering rhetoric spewed on Reddit as gospel. I'm just going to ask:

How on Earth does a power supply have nothing to do with how power is delivered and distributed? :confused:

Source of power gives you a possibility to draw energy, it is not pushing in any sense. Distribution is through the wires with the laws of physics. Read some elementary electrician books.

KenjiroThe tests with brand new connectors are useless. Try testing with old oxidised connector or socket with resistances on some pins even ten times higher than new. Problem arise with time when some pins have very different resistances than others.

I don't disagree, but hardware shouldn't be operated when it's old and oxidized. Also doesn't mean that it won't be; which is why safety is important.

I get it, arguing against an established standard like the (6+2) 8-pin PCIe connector from a reliability standpoint is an exercise in futility, it's an established, time proven standard after all. That's pretty much irrefutable, but it doesn't mean the new cable is evil incarnate either.

A quick look at an octopus adapter cable with 4 of those dangling around a RTX 5090 justifies it if you ask of me.

Just saw your edit: Bruh. Of course the GPU is drawing the power, except there isn't a single safety anywhere to cut this draw off in case of a malfunction on either side. That much should have been clear.

Dr. DroI don't disagree, but hardware shouldn't be operated when it's old and oxidized. Also doesn't mean that it won't be; which is why safety is important.

I get it, arguing against an established standard like the (6+2) 8-pin PCIe connector from a reliability standpoint is an exercise in futility, it's an established, time proven standard after all. That's pretty much irrefutable, but it doesn't mean the new cable is evil incarnate either.

A quick look at an octopus adapter cable with 4 of those dangling around a RTX 5090 justifies it if you ask of me.

Just saw your edit: Bruh. Of course the GPU is drawing the power, except there isn't a single safety anywhere to cut this draw off in case of a malfunction on either side. That much should have been clear.

PSU is not responsible for wrong power distribution through wires, it is the receiver side to be designed properly.
12V-2x6 connector is not designed to be immune to resistance differences on pins and VGA cards do not have any load balancing detection capabilities between pins.
Old 8-pin connector is almost immune to resistance differences on pins because of multiple pins with very little (relatively) current flowing through any of that pins (when any of that pins could safely take full current of the connector!) and VGA boards have load balancing detection between sockets.

This price for a thermistor XD

Dr. DroThe only thing that has been proven time and time again is that 2x6 has greatly improved the contacts, and that this isn't an Nvidia issue (and I'm tired of people pretending it is). The first few documented cases of Radeons that blew up with the connector are already in our hands. For example, the guy whose Taichi 9070 XT caught fire was running an underdimensioned 700 W power supply.

It's not the PSU. PSUs are dumb supplies, they will suppyl the amperage a device requests, they can detect shorts but they do not do handshakes with the linked device.

It is ont he device to manage its own power draw. This is the case with most electronics. The card makers need to be putting power resistors on the cards to force equal loading on all 6 pins. This was normal with multiple 8 pin connectors, but now card makers are cheapening the designs.

That is an nvidia, and AMD, and asrock, and asus, ece issue. It is not on the PSU. Card makers need to fix this.

Dr. DroIt doesn't take much of a genius to figure it out, the moment anything goes wrong, current follows the path of least resistance.

Apparently it does, since the internet still incorrectly labels this is a connector or PSU issue when debauer already debunked this as being a load balancing issue.

KenjiroPSU is not responsible for wrong power distribution through wires, it is the receiver side to be designed properly.
12V-2x6 connector is not designed to be immune to resistance differences on pins and VGA cards do not have any load balancing detection capabilities between pins.
Old 8-pin connector is almost immune to resistance differences on pins because of multiple pins with very little (relatively) current flowing through any of that pins (when any of that pins could safely take full current of the connector!) and VGA boards have load balancing detection between sockets.

It isnt immune. There were plenty of instances of 8 pins melting too, but since that wasnt pushed by nVidia, nobody cared.