Nvidia's RTX 3090 has been pictured - a 3-slot behemoth
- Thread starter WYP
- Start date
darthgalactis
New member
I've had an AX760 (non-i) for the last 5 years, and after I saw the recommended 850W rating I just bought the AX850 Titanium and Cablemod Mod flex kit. With the state the world is in right now it may be impossible to find one after the September 1st announcement. I also have a Meshify C and might even need to change my case if the dimensions for the 3090 turn out to be true. One the bright side the PSU will be a future proof investment so I'm not too concerned, but I'll be super annoyed if the dimensions mean I need to buy a new case.
AlienALX
Well-known member
I really don't know. No one does yet, at least where the 3090 is concerned.
What I will say is getting rid of the sort of heat that the new power connector could potentially supply will not be easy. A 400w GPU will need 400w worth of cooling. If the cooling isn't sufficient you end up with another Vega. Which was a total nightmare, even the massive Devil I had.
Personally I reckon it will have to be about the size of our cards (the Aorus you have and the Kingpin I have, or the triple fan MSI). Because those can use lots of power too, so have cooling to suffice. I just don't see how this new wonder cooler can deal with that sort of power draw without being absolutely huge. Which it is already, I mean, the whole thing is a bloody heatsink lol.
If I recall correctly, Gigabyte had a triple fan GPU that claimed to remove 400W heat, so if thats true, it could be the cards dont change much in size. But I was talking more about length and width. It could possible end up being much deeper allowing for deeper vapo chamber.
Well, NVidia have cooled 375W cards with blowers before...
But I doubt the new connector is specifically intended for raising the power limits(Especially given just using 3x8-pins would have been a much easier solution if only targeting that single issue), this connector clearly has overall benefits to unification of GPU connectors(Assuming further adoption) and board space, besides of course increasing the number of pins/conduction area having overall benefits to transmission efficiency, reducing voltage drop, reducing connector & wire heat up, and allowing much more redundancy for power bursts/spikes or heavy overclocking(By reducing overall line connection/line resistance ofc). Much the same way we see some high end cards go with 3x8-pins even with <300W stock GPUs, there are more benefits besides just raising the power limit.
But I doubt the new connector is specifically intended for raising the power limits(Especially given just using 3x8-pins would have been a much easier solution if only targeting that single issue), this connector clearly has overall benefits to unification of GPU connectors(Assuming further adoption) and board space, besides of course increasing the number of pins/conduction area having overall benefits to transmission efficiency, reducing voltage drop, reducing connector & wire heat up, and allowing much more redundancy for power bursts/spikes or heavy overclocking(By reducing overall line connection/line resistance ofc). Much the same way we see some high end cards go with 3x8-pins even with <300W stock GPUs, there are more benefits besides just raising the power limit.
looz
Active member
Gotta remember that Vega had issues in cooler design - mounting pressure was all over the place and HBM wasn't flush with the die. And the die itself was a significantly smaller than Turing for instance.
400W in itself isn't unfeasible to get rid of, when it's a large die with direct cooling. But I hope it's going to be more efficient than that.
DwarfAdigginAhole
New member
OMG come on. That's so fake it's ridiculous. It's that same fake crap from before, crudely enlarged so that the PCI slot on the "3090" is bigger than the 2080ti (1), fans that are so wrong to be in this type of design and one with horrile defects (2) and why can I see through the fins to the table in the middle of the card , where there should be pcb and backplate as per the other picture (3) and why can i see through the IO? (4)
This has been confirmed by many people actually in the industry to be a legitimate leak. Nearly everyone has picked up on this one bud. This isn't a fake lol.
Dark NighT
New member
As I said, "IF" this is legit, we are looking at early 3d Printed models and no where near the final design. There is no way on earth what we see is a legitimate cooler for the 3090 as the scaling is off, fins/vanes and overall design looks very alpha/photoshopped.
Post a source if its been confirmed, otherwise I stand by my original opinions.
AlienALX
Well-known member
If it is legit then it is out of whack completely. And out of scale, and out of perspective.
It may have a shoulder and it may be bigger than the 3080 we have seen but that's about it. IE, that cooler may be real but the back plate and PCB some one added on are fake (they don't fall into PCIE spec at all) and they have placed a PCB underneath it and photoshopped in the back plate.
The cooler may be a prototype (because that one appears to be very crude and 3d printed) but the rest? doesnt add up and that size scale thing is a joke. If it is that much bigger than the 2080Ti it won't fit in any cases.
It may have a shoulder and it may be bigger than the 3080 we have seen but that's about it. IE, that cooler may be real but the back plate and PCB some one added on are fake (they don't fall into PCIE spec at all) and they have placed a PCB underneath it and photoshopped in the back plate.
The cooler may be a prototype (because that one appears to be very crude and 3d printed) but the rest? doesnt add up and that size scale thing is a joke. If it is that much bigger than the 2080Ti it won't fit in any cases.
I think the size/scale issues you're perceiving may be because your comparison/skew compares a traditional image at an off-angle with an essentially orthographic image of an RTX2080Ti.
With a bit of geometry someone could get some fairly accurate estimates for the dimension in this image, think that would be much more reliable and provable way to go about it than image skews, if someone has some time to burn and wanted to try it
With a bit of geometry someone could get some fairly accurate estimates for the dimension in this image, think that would be much more reliable and provable way to go about it than image skews, if someone has some time to burn and wanted to try it
Last edited:
AlienALX
Well-known member
Another site lined up the PCIE slots. However, they failed to take into account the height of the slot. So once again it was inaccurate.
That may well be the cooler, and it may well have a shoulder (like all of the big boy cards, HOF, Kingpin, Duke etc) but it ain't as big as they are trying to make out.
Because if it is the fans are 150mm and the height of the card will be about 160mm at least which means nearly all cases won't close.
The reason I am so adamant about that is because I now create all of my modding templates (for hand cutting acrylic and metal) in Photoshop. Mostly because it does point mm (So for example 26.8mm) and I know for a fact that when I send that image to my plotter and cut stencils that I won't have things fall out because the holes are too big, or things not fit because they are too small. It saves me hours of filing and waste materials.
So because of that I have spent a lot of time drawing scaled objects. So it's one of a few things. Fake? wrong perspective? overblown and much larger than it really is? thrown on a work bench and placed onto the PCB and so on. But it isn't right.
As I say, I fully expect it to be a chunk bigger than the 3080 but not to the point you can only run it on a test bench with the PCI bracket removed because it sticks out of the back of the case by over an inch. Nvidia would not be that stupid.
Similar threads
