Here is an alleged photo of AMD Radeon Big Navi Radeon RX 6000
Click here to post a comment for Here is an alleged photo of AMD Radeon Big Navi Radeon RX 6000 on our message forum
Devid
Making a 256bit card to compete with what exactly? 😱
There is no way this the full fat Big Navi maybe this is its little sibling...
Undying
It seems fake and somebody is playing with his 5700xt.
Netherwind
Still think this is too late. Wonder why AMD decided to pick up the scraps after the release of the 3000-series. I guess nVidia must be pretty happy.
kegastaMmer
"too" late is a bit too early, since we have no cards on hand, no reviews in sight. And if we wanna guess how NVIDIA is feeling, i'd say they shit their pants when they obtained some inside info about pricings of RDNA2 cards which is why they lowered their prices, out of panic, out of sheer desperation after decades of price gouging.
Undying
If first 3080 benchmarks are true nvidia will be not happy for long.
Kaarme
It's not like AMD, which was a part of the HBM design team, wouldn't know the significance of memory bandwidth. Why would they suddenly starve their cards on purpose? Unless they came up with some phenomenal, previously unimaginable way to double the bandwidth use efficiency.
WhiteLightning
Moderator
Is this codename "Dust magnet" ?
Venix
Yeah this rig shows way too much dust ...one thing is certain this pc on the picture needs some love !
suty455
Would say that was a small er Navi 🙂 the dust could be deliberate lets face it not everyone cares how clean the inside of a pc is
Netherwind
The leaked ones from china? 😀 I hope they ain't true or 2080Ti users won't have anything to upgrade (unless they pay extremely much for 3090)
Astyanax
its not fake, the stickers on the card designates it as "Full ???? Typical XT A0 ASIC’"
itpro
If you want some real inside info to play with,
Go search videos and benchmarks. The whole situation will be like 280/280x/290/290x vs 760/770/780/780ti all over again. Nvidia will not be bang for buck and will not hold the performance crown so easily , unless they out a real titan.
200 series, either x or normal, were and still is kinda undead gpu. Like gtx 900 series vs r9 300 series.
Future-proof, better and more memory. Both r9 200/300 series had more memory bit, bandwidth, features than 700/900 gtx series, 3/4/8 gb vs 3/4/6 gb.
Truder
If you look at the back of the board, you can see various pins and sockets, probably diagnostic pins
Silva
I don't see this card beating 3080 let alone competing with 3090, you guys need a reality check.
Should be 2x 5700XT performance plus something we have no idea of how much.
Well, end of next month we will know.
Fediuld
It is fake. A testboard true but not from the upcoming cards.
This guy is nobody and has access to that card in China of all countries?
Clearly reminds me the Vega leaked picture from June 2017 tbh
JamesSneed
If its real(I don't believe it is) its probably Navi22 which is the bigger but not biggest die. There is no way the largest Navi die will have a 256bit memory controler.
Undying
If its real and its second best navi its strange that it has 16gb. Not that i complain though. It spells doom for 3070.
Evildead666
I'm going with the 'Bits and Bytes' theory, that this is 16Gbit chips of Samsung GDDR6 being shiowcased, along with the A0 rev of the card.
It looks like those stickers were put there to show off the components to people, maybe during a backroom demo a long while ago ?
If Nvidia can keep their cards under wraps until launch day, I suspect AMD can too. 😉
(I'm also secretly hoping that the differentiation between PRO and Consumer GPU's at AMD will enable the removal of Double Precision hardware from the consumer side, and increase the number of 'simpler' shader units. Remember what happened between the Radeon 2xxx series and the 3xxx series ? a large jump in shader count.
Could this be why Nvidia upped the shader count so much ? because of a possible 80CU / 128 shaders per CU card at AMD ?)
sbacchetta
At the same time a 3080 has the same number of ROP, TMU, Int ALU and similar clock speed than a 2080ti. In the other hand, yes, the Ampere card has 2x the FP ALU and more memory bandwidth, but unless you look at cases where those were serious bottleneck for a 2080ti, the improvement won't be massive.
Nvidia is paying the price of Samsung 8nm and zero improvement to clock speed....
Shadowdane
Well if it actually has 16GB it's going to be 256-bit memory controller, no other way that will actually work using an 8 x 2GB memory configuration. 8 x 32-bit memory controllers = 256-bit. The # of memory controllers has to match the # of chips on the PCB. If they do GDDR6X same as the RTX 3080, it would have 608GB/s using 19Gbps GDDR6X memory.
Short of them doing 16 x 1GB configuration which I also doubt as that would make for a very crowded PCB layout. Also kinda rules out them putting them on the back of the PCB too as I don't think the backplate would provide enough cooling for GDDR6 or GDDR6X. Not to mention a 16 x 1GB memory config would end up being a 512-bit memory bus.
