AMD Radeon RX 7900 XTX RDNA3 Prototype Leaked, Confirms Reference Cooler Design

[btarunr]

Here's what is possibly the very first picture of an AMD Radeon RX 7900 XTX RDNA3 graphics card. AMD engineering samples and prototypes tend to use red color PCBs, which is what this card is. It reveals what could be the final design of the reference cooling solution for the card, and it seems to match the teasers the company put out in its Ryzen 7000-series launch event.

The RX 7900 XTX cooling solution design builds on that of its predecessor. The card itself has 3 slots thick, but slightly longer than the RX 6900 XT. The aluminium fin-stack heatsink is bulkier than the one on the RX 6900 XT cooler, and appears to be bursting out of the vents. It stretches out to the edges of the cooler shroud. The bulge toward the tail-end could be housing the tips of the heat-pipes. The prototype card has two 8-pin PCIe power inputs. There's no backplate, because the PCB has several headers in place for diagnostics and developmental use by AIBs and OEMs.



View at TechPowerUp Main Site | Source

 

[MachineLearning]

Can only speak to aesthetics, maybe it performs better.
New design is ugly AF.

 

[Minus Infinity]

Yeah, I wouldn't care if a GPU looked like a dog turd, I never look at the inside of my case except for it's 3 monthly cleaning.

 

[Dirt Chip]

It must be the most efficient GPU ever, to match 4090 preformance with only 300-325w (300 from 2*8pin for stock use + 50-75w as reserve from pcie).
Well done AMD!

 

[uftfa]

2 power connectors? So, the 400+ TBP claims are not true for the reference cards? Doesn't give me much hope for it competing with the 4090 even in rasterization then.

It must be the most efficient GPU ever, to match 4090 preformance with only 300-325w (300 from 2*8pin for stock use + 50-75w as reserve from pcie).
Well done AMD!

I don't think AMD will get anywhere close to that level of efficiency, but to be fair, benchmarks of 4090 with 300W power limit gets about 91% of the performance relative to "full" performance at 450W. If AMD has 10% efficiency advantage, they'd be able to match 4090 full-power with 300W.

However, AMD is slightly behind nvidia in terms of fab process this gen, and MCM will have at least a small performance penalty compared to monolithic, I simply don't see the winning on pure efficiency this gen.

 

[Zubasa]

2 power connectors? So, the 400+ TBP claims are not true for the reference cards? Doesn't give me much hope for it competing with the 4090 even in rasterization then.

The 4090 doesn't run into its PL in raster, it only does so in RT heavy games.

 

[ZoneDymo]

looks solid to me, but as GN showed, its more about the contact with the chip that was a bit lacking on the old design.
I like how the triangles on the fanhub visually connect the fins, plays with my eyes a bit, its of no consequences, but its the sorta thing I would do myself

 

[john_]

With AMD going dual 8pins I can thing of two things. First, power consumption will be lower than 350W. And, on the contrary to Nvidia, AMD is not trying to compete with it's AIBs. AIBs can probably design their cards to have three 8 pins and offer higher frequencies.

2 power connectors? So, the 400+ TBP claims are not true for the reference cards? Doesn't give me much hope for it competing with the 4090 even in rasterization then.


I don't think AMD will get anywhere close to that level of efficiency, but to be fair, benchmarks of 4090 with 300W power limit gets about 91% of the performance relative to "full" performance at 450W. If AMD has 10% efficiency advantage, they'd be able to match 4090 full-power with 300W.

However, AMD is slightly behind nvidia in terms of fab process this gen, and MCM will have at least a small performance penalty compared to monolithic, I simply don't see the winning on pure efficiency this gen.

In rasterization might be even faster, but it will be a success for AMD if at least beats RTX 3000 in RayTracing.
Nvidia's GPUs are more complex than AMD's meaning they also include Tensor cores. AMD split it's gaming and server design, so yes, it can be more efficient in gaming. But probably it will still get butchered in AI and pro applications.

 

[MarsM4N]

I like the fins, esp. the 3 red ones (RDNA3). Other than that, too much plastic. I hate plastic.

2 power connectors? So, the 400+ TBP claims are not true for the reference cards? Doesn't give me much hope for it competing with the 4090 even in rasterization then.

wccftech reported that the 7900XT will be 350W, so that's enough for 2x 8pin + 50W PCIe power. Does AMD have to compete in the rat race of melting power cables?
I don't think so. Most people will be happy with 95% performance for 100W less power & heat.

 

[igralec84]

Damn, December can't come soon enough!

 

[uftfa]

wccftech reported that the 7900XT will be 350W, so that's enough for 2x 8pin + 50W PCIe power.

Yes, but this is the 7900XTX. Would be a bit surprising if the XTX has the same TBP as the XT. Although there is historical precedent for it going back to the R9 290 and 290X

 

[TheDeeGee]

It must be the most efficient GPU ever, to match 4090 preformance with only 300-325w (300 from 2*8pin for stock use + 50-75w as reserve from pcie).
Well done AMD!

8-Pin is rated for 342 Watts

 

[junglist724]

The 4090 doesn't run into its PL in raster, it only does so in RT heavy games.

I can only get my 4090 to exceed 450W in some rasterized games, RT heavy games often run into an RT core bottleneck or CPU bottleneck. In something like Control my power consumption goes up when turning RT off.

 

[ZetZet]

However, AMD is slightly behind nvidia in terms of fab process this gen, and MCM will have at least a small performance penalty compared to monolithic, I simply don't see the winning on pure efficiency this gen.

That's bullshit. N4 is pure marketing...

 

[Arkz]

8-Pin is rated for 342 Watts

Wat? 8pin is rated 150w

 

[Crackong]

8-Pin is rated for 342 Watts

The PCIE 8-pin power is rated 150W for PCIE spec.

But the actual capability of those things can go up to 360W per connector if using HCS rated components.

 

[BlaezaLite]

We know what it looks like now, just performance and the dreaded price. Please let AMD have some braincells and do top of the stack for £999.

 

[Punkenjoy]

However, AMD is slightly behind nvidia in terms of fab process this gen, and MCM will have at least a small performance penalty compared to monolithic, I simply don't see the winning on pure efficiency this gen.

Both AMD and Nvidia use the same "Base Process" this gen. They each have a custom process out of it. The 4nm is only in the name and it's marketing. Process stopped matching nanometer when they switched to 14nm/finFET, if not before.

These days, transistors size only really decrease between transistors technology. We will see smaller transistors when they switch to one of the gate all round tech. But before that all of those are just tech to pack a similar size transistors as tight as possible.

There is still many benefits of that, the main is the circuits became shorter while doing it meaning less energy spend propagating signal and since electron have less to travel, it's easier to clock thing higher. But in the end you need about the same power to make the transistors switch.

The mains reasons why AMD might be so efficient this gen is 1, they still probably don't have anything like the tensor cores that are dark silicon if you don't use DLSS and 2, they cleanup as much dark silicon from the past. Their new chip will be very lean.

They won't be able to redo that trick a second time and at some point, they will need tensor cores or something equivalent. But right now, they will be able to produce a much smaller chip that will be easier to power and cheaper to produce.

 

[Dimitriman]

I got a bad feeling about one thing... If we are getting both a 7900 XTX and 7900XT instead of 7900XT + 7800XT, I am not hopeful for the pricing. This is gonna be 4080 12GB all over again.

 

[Legacy-ZA]

I, uh, kinda like the design.

 

[ymdhis]

If you are making the card 3 slots high, you should design it so it can accept a 120x25mm fan. Those wimpy 80x12mm fans barely do the job.
Scythe used to have this monstrous 5-heatpipe gpu block that was 2 slots tall with a slim 120x12 fan, but you could put a 25mm one there (making it 2.5 slot tall). It made the card silent and have significantly better temps than the crap they put even on modern cards. It was only 200W though, but it was also single fan...

 

[spnidel]

new design looks great

 

[watzupken]

I like the fins, esp. the 3 red ones (RDNA3). Other than that, too much plastic. I hate plastic.



wccftech reported that the 7900XT will be 350W, so that's enough for 2x 8pin + 50W PCIe power. Does AMD have to compete in the rat race of melting power cables?
I don't think so. Most people will be happy with 95% performance for 100W less power & heat

PCI-e slot should be able to provide 75W at least. But with 2x 8pin, I believe it is capable of delivering more than 300W. Assuming its sub 400W while delivering good rasterization performance, I think it will be pretty decent alternative. I don't think AMD will catch up with Ada in terms of RT performance, but the fact that Nvidia had to cook up some frame inserting tech may prove that performance gap may not be that big between the 2 camps.

 

[jesdals]

Thats 20mm to long for Meshify C setup - damn

Btw @Wizzard could you include in your upcomming test if thermalpads on the back plate is needed like on the reference 6900/6950XT and how thick it need to bee.

 

[Arkz]

I think it looks good tbh, but I miss the big Team Rocket R on the fans.