giving it more thought, i dont think this connector is electrically backwards compatibile with pcie, but it is mechanically.
A normal gpu in an asus slot would work, but still require its pcie power connectors to be plugged in.
An asus gpu in a normal pcie slot would pull too much current and damage the mobo.

john_Because we would love 850W graphics cards.

I would love more cards that need no or just 1 powercable, and if that brings that, cool.

For what those things cost today they should just add a power brick in the box and have the cards get 48V from that

Great idea on paper bad in use imo.

Let have MORE wattage going threw MB then we already have with CPU let now add GPU power too it too, those we need really beefy MB PCB layers among other thing which drive prices even higher then stupid they are now. But now one can have there MB and probably CPU and everything else go at same time as gpu.

Leave psu to feed the gpu and nvidia and amd need to put proper management of power draw on board/gpu them self so they dont do stupid thing. and stop using the 12x6 connector.

I like the idea of single cable or even none, but not expensive safety

copper on motherboards is very inadequate for high power. The CPU power connector is immediately near the regulators. Even that has limitations which keep CPU power consumption in check.

Power pig video cards have issues with feeding them 600W or more. PCIe cables are not designed for such extremes. 6-pin 75W and 8-pin 150W is the extent of PCIe power standards.
Radeon card I had needed 3x 8-pin cables.

OnasiTo add to that, even high end GPUs are over-specced in the power department and don’t actually need to be THIS power hungry. In the recent thread, @wolf ran tests with limiitng and tweaking his 5080 and running a heavy GPU load and the consensus was that the 5080 can easily be a 250-280W card. The extra power does nigh on nothing.

Same with 4090 - 320w is where it's at.

Chrispy_but now you need a new motherboard, too....

windwhirl... To be honest, I'd feel this whole move would be just a justification to increase motherboard pricing even more.

And WHO makes mobo's, and would therefore benefit from the higher pricing.... ?

Yep, you guessed it....AsSus (among others) !

And the sad part is that if even 1 of the other major mobo makers adopt this, it will become widespread, and therefore considered a "standard" faster than flies can locate & land on a pile of fresh, steaming hot poop...then we'll all essentially be forced to accept it, along with the correspondingly higher prices :(

Nope, not doing that, and I'll just stick with my little, powerful-enough-for-me mini-me boxes, and be happier (and richer) for it... :D !

To be honest if we wanted to fix this problem for good, we'd need a completely new motherboard form factor.

ymdhisTo be honest if we wanted to fix this problem for good, we'd need a completely new motherboard form factor.

We can't even get rid of 3.3V and 5V. Or M.2 form factor for desktops, for that matter.

JustBenchingSame with 4090 - 320w is where it's at.

My 6950XT gains around 11% core clock with a near 100 watt increase. (270 to 365)

Not worth it, my card sits at a -10%/270watt max limit 24/7

Is ASUS in that desperate of attention that it figured out it has to propose more points of failure?
Le: it also waves hello to nVidia's make believe power connector that is way more complex,but by the power of make believe also cheaper.

freeagentThis would help prevent 12VHPWR plugs from melting I think. I know for sure they can take 400w :D

Until Nvidia release 850W 6090. I think they would be too tempted to. :)

Tek-CheckNot sure why it is a big deal to oppose this solution. Let them release a few retail boards to experiment with uptake, and whoever wants to buy it, be my guest. It's about having choices. By introducing a few boards with this solution, users can still connect PSU directly to appropriate GPU, or via motherboard too. If someone wants to pay more for a board with such option to improve aesthetics of their build and hide cables, it's their choice.

The first thing, that would be a problem is the industry would want to standardise, rather than having some users having a 75w PCIe and others 250w. So it would eventually not be a choice.

Then you would have the problem where you might want a board, that is definitely not budget category, so you pay extra for nice i/o options, overclocking chipset, but you have no interest in paying for a luxurious PCB that can feed 100s of watts to a PCIe slot. You might also not want the added point of failure, however the board vendors choice might be you pay for it, or you buy a bottom of the barrel $100 board without the feature and all the downsides with that, so instead of choosing between a $100 board or $400 board its now $100 or $1000.

Currently you buy a good PSU and you reasonably assured of good power delivery for a decade or so, this would change all that.

This is a great idea! Not the first time someone has tried this, but still a step in the right direction.

sudothelinuxwizardIs it just me or does pushing 225W through less then 10 microscopically thin copper traces sound like a fire hazard?

This comment makes absolutely no sense at all. If the connectors are rated for and stay at or under the rated value then it doesn't matter how many of them there are and what the total is as long as the PSU can supply it. Where's your critical thinking? Out the window.

WirkoWe can't even get rid of 3.3V and 5V. Or M.2 form factor for desktops, for that matter.

No one really needs 5V DC for USB, right? A few weeks ago I quick checked those uefi chips: no one really needs 3.3V DC right?

Basically you are on a mainboard which generates those voltages from other voltages which just will increase the mainboard price. I'm really surprised on the 5V one. Any Smartphone charger, usb keyboard, usb mouse shows that Voltage instantly on any label.

--

Power supply units needs to handle spikes. Much more complicated as the fancy big 750W number.

--

Asus should learn the basics about I/O lines and power delivery.

chrcolukThe first thing, that would be a problem is the industry would want to standardise, rather than having some users having a 75w PCIe and others 250w. So it would eventually not be a cho

There's no evidence that such solution would be standardised. Even power cables are still diverse, one type usually used by Nvidia and another one by Intel and AMD cards.

chrcolukYou might also not want the added point of failure, however the board vendors choice might be you pay for it, or you buy a bottom of the barrel $100 board without the feature and all the downsides with that, so instead of choosing between a $100 board or $400 board its now $100 or $1000.

This is extremely pessimistic scenario assuming the industry has an appetite to standardise the solution and offer it across the board. Again, there is no evidence of such approach. The best case scenario is that a few boards would have it, to test waters.

freeagentThis would help prevent 12VHPWR plugs from melting I think. I know for sure they can take 400w :D

Technically 12VHPWR is only safe to a maximum of 100W, as that is what each pin is rated for. And since all cards are designed in such a way (because NVIDIA engineers) that 600W can flow through 1 pin, the maximum rating for 12VHPWR should be 100W.

ZoneDymoI would love more cards that need no or just 1 powercable, and if that brings that, cool.

We will get cards that will not need a power connector but at the same time we will get incompatibility with every motherboard in the market. It will take years for the new standard to prevail and the new PCIe slots will need to have a different shape to make sure that people will not be putting PCIe cards that expect to get 250W from a current PCIe slot, that can only offer 75W. I think ASUS's idea that is described in this article is a stupid idea that will lead to problems, while it doesn't really makes the setup less of a cable mess. Instead of connecting a power cable on the back of the graphics card, you have to connect another on one side of the motherboard with the rest of the power cables connected to the motherboard being on the other side of the motherboard. Not to mention that connecting a power cable on the graphics card could lead to a burned graphics card. Driving the power through the motherboard could lead to a burned graphics card AND a burned motherboard if something goes wrong.
Where they drunk when thinking of this?

Long story short, this idea adds incompatibility and many opportunities for the average Joe to burn his system or have no idea why his new graphics card doesn't work on his not so old motherboard. And I believe this is more of an idea to push power consumption of graphics cards over 675W, than just offer "cable less" options up to 250W.

Did they tested this properly?
I would loved to see a normal production mb like this (not a sample that they presented and most likely they sent to reviewers), that had the pci-e slot abused (in the sense that a gpu has been plugged and removed in it for a lot of times, close to the limit of use for a pci-e), used with a gpu that also was abused same as the pci-e (same conditions as in the pci-e slot case, this gpu can be used to abuse that pci-e slot so the numbers match) run 24/24 7/7 for an entire year drawing 225 W (just 90%). I have my doubts that such an experiment will end up well, despite the fact that the theory saw that it should work totaly fine.
The higher the power drawn on an warn connector, the higher the probability of problems.

Wait until someone will say "We need to reduce the production cost to increase the profit." and they will cut the material used. It's always like this...

Joking: If this will go into production soon we will have motherboards with smoke detector and fire extinguisher 🤣
Maybe the case manufacturers will be the first ones to add the smoke dector and fire extinguisher.

If Asus will be the only manufacturer implementing this, I'm totaly fine with this, they can have their own ecosystem (they can have their asus mb and produce the asus gpu just for their mb). I blacklisted Asus long time ago for no customer support, extremely bad quality of bios (was just compiled and because it compiled fine they assume it also works fine, stayed broken from the version the problem showed up until EOL, only solution was to stay with the last bios that didn't experience the problem) + vbios, insane customer support agent answers (he admited that the costs to send it to their service located in another country + cost of flashing it are higher than a new product but he went on writing the same logical abberation also refused to escalate the case as I requested, hard to send me what I asked in the ticket...). I don't care they are now top class with the customer support (I doubt), when I blacklist a manufacturer (I need more than 1 case) the blacklist also comes with "DO NOT REEVALUATE" flag.

TheDeeGeeTechnically 12VHPWR is only safe to a maximum of 100W, as that is what each pin is rated for. And since all cards are designed in such a way (because NVIDIA engineers) that 600W can flow through 1 pin, the maximum rating for 12VHPWR should be 100W.

I guess my card should have self destructed when I plugged it in 3 years ago.

I have been saying this for a while now (a few years), the whole ATX standard needs to be scrapped and we need a new one.

ATX was introduced in 1995, that's 30 years ago in a time there was no PCIe, even graphics cards were not known as GPUs, they were called graphics accelerators, they didn't even had any fans and consumed maximum of 15W. And the CPU was even lower at 10W.

Now these 10-15W only is taken by our lovely RGB lightings (if not more) and another maybe x2-4 times that just to cool down the system which can take 500W+ full load. CPUs and GPUs went from small un cooled, to a heatsink then a fan and becomes larger and larger everytime as we require more performance that what the process technology advance allows us to have. The way we install GPUs & M.2 drives takes up a lot of space of the motherboard which is inefficient, thats why we have 3+ slots GPUs these days and full ATX motherboards that only have 2x PCIe slots because M.2 slots takes all other space.

TheDeeGeeTechnically 12VHPWR is only safe to a maximum of 100W, as that is what each pin is rated for. And since all cards are designed in such a way (because NVIDIA engineers) that 600W can flow through 1 pin, the maximum rating for 12VHPWR should be 100W.

That's an interesting way of puting it, I'd rather say 12VHPWR is just not safe.

Millions of users may be fine, doesn't mean it's not a stupid design that prioritises space efficiency - that doesn't matter when attached to a brick of a GPU - over of power capacity and safety.

freeagentI guess my card should have self destructed when I plugged it in 3 years ago.

Only if contacts are bad or you physically cut wires.

Doesn't take away that the design of the power delivery is fucked up, and has no safety systems in check.

Safest card would be the Astral which monitors the pins, though you still have to take actions yourself to prevent a disaster.

trsttteThat's an interesting way of puting it, I'd rather say 12VHPWR is just not safe.

Millions of users may be fine, doesn't mean it's not a stupid design that prioritises space efficiency - that doesn't matter when attached to a brick of a GPU - over of power capacity and safety.

The connector is technically safe, though the safety margin is very very small on a 600W card. The pins are rated for 9.5 Amps, which is in total givenor take 625W for the entire connector. This connector should be rated for 300W, but only if the power delivery on the board is balanced.

That said AMD cards with the same connector have this issue as well. The Sapphire card has the same poor board design which allows more than 100W per pin. Though i don't believe AMD forces these strict rules and partners have more freedom, but sadly some of them still don't bother.

It's all about saving pennies.

TheDeeGeeTechnically 12VHPWR is only safe to a maximum of 100W, as that is what each pin is rated for. And since all cards are designed in such a way (because NVIDIA engineers) that 600W can flow through 1 pin, the maximum rating for 12VHPWR should be 100W.

Following the same logic 8 pin PCI-e should be limited to 50 watts then as thats is the "safe" limit of the spec through a single conductor


The problem is when a standard is designed with certain "expectectations" to be followed which are then removed by bean counters to gain a few cents per card when it should have been made part of the base specification.

Panther_SeraphinThe problem is when a standard is designed with certain "expectectations" to be followed which are then removed by bean counters to gain a few cents per card when it should have been made part of the base specification.

big part problem.

Nvidia use be balance that stuff and even made point state they balanced they then stop and AMD followed suit. Most issue are stemming from them fact load is NOT balanced anymore