I don't get it. Two different reference cards of the same card, yet not the same card, and for some reason a much smaller PCB, yet not a smaller card? None of this makes sense.

much smaller PCB, yet not a smaller card?

If the card requires this level of cooling, can't just cut-off a third of the heatsink. 🙂

Maybe it's some kind of a demo board. It has no backplate. What are those three connectors directly on the PCB anyway? Do normal video cards have such under the backplate? I'd say no.

The pcb is half the size of the cooler... either they went overboard or we should we worried if it need some that much cooling... hopefully not the latter... Nice to see custom cards are going to come out anyway, hopefully we might see a new nano card for those who like to build small itx machines

The reason the card is so long is because for some people, the bigger the card, the MOAR performance it gets. i call them Rice GPU's. So remember kids, never buy a gpu if its small, go for the BIGGEST one, cos it performs better.

Looks very similar to my Sapphire Nitro R9 Fury, except this one has AMD branding instead of Sapphire. Also, to everyone in the comment section who doesn't understand the point of putting a big heat-sink on a short PCB: Big heat-sink equals good cooling performance at low noise levels. The short PCB actually is advantageous for air flow and increases cooling performance even more than it would be with a long PCB, which blocks the air flow more. Pretty simple really. Don't get why so many have trouble to understand...

Indeed its not really custom. Its made it by Sapphire(who made all Firepro cards) for AMD.Still reference card. 😉 Maybe will see in future a RX Vega Nano card.

This kind of heatsink is for people like me 🤓 I personally absolutely don't mind it, in my case, big coolers served me better than sleek cooling solutions. Used to have Palit 970 with small PCB but with an extended cooler and it was one of the quietest and best performing cards on the market. Now I using TRI-TRI-TRIPLE SLOT 1080, absolutely massive, but it fits the case nicely and is as silent after running for hours as it can get.

The reason the card is so long is because for some people, the bigger the card, the MOAR performance it gets. i call them Rice GPU's. So remember kids, never buy a gpu if its small, go for the BIGGEST one, cos it performs better.

Sounds like sarcasm but is true in this case 🙂 BIGGEST cooler to the very least will mean lower temps = better potential performance for longer sessions and lower overall noise.

Now that´s a decent cooler and looks very nice too.

That would be the first graphic card with scaffolding

That would be the first graphic card with scaffolding

Nope

This is possibly one of the Very Best designs for a GPU IMHO. I have been Really hoping for a Vega version of my Sapphire Fury Tri-X: http://us.dl.sapphiretech.com/archive/gm/PDimg/Sapphire_201526165319.jpg The last fan having no PCB behind the heatsink gets fantastic airflow for great cooling requiring lower fan speeds. In fact this cards fans power off when below about 75c. (This saves the fans from wearing out while not gaming etc - had lots of GPU fans die in the past 🙁 ) In fact in my PC case with good airflow the fans almost Never spin up even in 4k gaming, even when mining coins they only switch on in hot weather!!!

These I like, I bet they are the XFX versions. 🙂 Replacing my 2 x Furies non X with 2 of these would be a beautiful thing. 🙂 (Not going too but, it would be nice.)

finally we have a nice PCB. Let's make a dual fan nitro+ with backplate now...

If the card requires this level of cooling, can't just cut-off a third of the heatsink. 🙂

That's not the point. The point is, if you're going to make a smaller card, what's the point if you're not going to actually MAKE it smaller, in the end? Unless they expect people to use water cooling on these cards.

The short PCB actually is advantageous for air flow and increases cooling performance even more than it would be with a long PCB, which blocks the air flow more. Pretty simple really. Don't get why so many have trouble to understand...

...no, if you're going to make a claim like that, you're going to need proof. If it were that simple, which is guaranteed to not be that simple, but if it were, then every GPU manufacturer would always make their PCBs small, because why spend extra money on more PCB when in reality, if it were smaller, you'd get a better performing card (temperature), with a lower cost, selling for the same price? This is how business' work, you can't just say "Oh well they just don't think about it", as that's nonsense. ANYTHING to save a buck and make your product BETTER is exactly what they would do. I understand what you're saying, and i understand why you would think it, but the reality is, unless there's a review that shows you are correct, you plain and simple can't make statements like that as if you are correct without proof. Reality states you are wrong, due to my above statement. And before someone goes and looks at temperatures of the fury card listed above vs other fury cards: The only way to to a correct review to make a statement such as this something that is correct, is to use the exact same cooler, on both cards. You can't compare a small PCB with a 3-fan cooler vs a large(or regular) PCB with a different 3-fan cooler. That's not how testing works. You would have to get the exact same GPU, but one with a small PCB, one with a normal PCB, running at the exact same specifications and case setup and etc., and run them both with the exact same cooler.

That's not the point. The point is, if you're going to make a smaller card, what's the point if you're not going to actually MAKE it smaller, in the end? Unless they expect people to use water cooling on these cards. ...no, if you're going to make a claim like that, you're going to need proof. If it were that simple, which is guaranteed to not be that simple, but if it were, then every GPU manufacturer would always make their PCBs small, because why spend extra money on more PCB when in reality, if it were smaller, you'd get a better performing card (temperature), with a lower cost, selling for the same price? This is how business' work, you can't just say "Oh well they just don't think about it", as that's nonsense. ANYTHING to save a buck and make your product BETTER is exactly what they would do. I understand what you're saying, and i understand why you would think it, but the reality is, unless there's a review that shows you are correct, you plain and simple can't make statements like that as if you are correct without proof. Reality states you are wrong, due to my above statement. And before someone goes and looks at temperatures of the fury card listed above vs other fury cards: The only way to to a correct review to make a statement such as this something that is correct, is to use the exact same cooler, on both cards. You can't compare a small PCB with a 3-fan cooler vs a large(or regular) PCB with a different 3-fan cooler. That's not how testing works. You would have to get the exact same GPU, but one with a small PCB, one with a normal PCB, running at the exact same specifications and case setup and etc., and run them both with the exact same cooler.

Not that I see much point in arguing with you, but you seem to forget a few things. Most cards don't have HBM, so they need extra space for the memory modules and thus have to move the VRM further back. On a Fury, or Vega, there's really no point in doing that, since you have plenty of space to put a huge VRM-array where the memory modules usually sit. Actually, you want to have the VRM as close to the GPU as possible for getting better efficiency and transient response, so there's no need to extend the card to make it longer, as that area of the PCB would be completely empty anyway. And yes, regarding air-flow, of course it's that simple, why wouldn't it be? Try to stick a piece of cardboard where the PCB ends to block airflow like it would be on a traditional card. Of course you would see temps and fan speed going up. There's a reason the Sapphire Fury's scored wins in most reviews in regards of noise and temps, despite being power hungry cards. Just go back and read the reviews. Not saying the short PCB is the only factor to the good performance of this particular heat-sink, but it's part of it.