Support NeoGAF

[ResilientBanana]

128GB of GDDR7 and only 4090 performance? Firstly these rumors are bullshit, secondly, you're looking at a $3000 GPU

 

[PlayStaytion]

This would mean that 5090 is 1.32x the performance of 4080. Are you possibly confusing 4080 with 4090?

Techpowerup sounds biased to me.

 

[PlayStaytion]

Trees in Wuchang were tanking my RTX 4070 again. Hope these new cards can speed up the trees.

I don't want a $3000 gpu that behave like a $500 gpu, we should've gotten over this bs long time ago.

 

[Gaiff]

128GB of GDDR7 and only 4090 performance? Firstly these rumors are bullshit, secondly, you're looking at a $3000 GPU

This just the max amount of memory for that particular architecture. You won't see any 128GB GDDR7 GPU just like you won't see 184 CUs. The top of the line consumer version will be 36GB with 154 CUs.

 

[Durin]

Honestly, for what though? At this point bandwidth is more important than the actual ram amount. Texture quality is already really good with 16GBs, so what is 36GBs necessary for? Its not like there is going to be this huge push to higher than 4K resolution anytime soon.

I could see moving to 24GBs so you have more headroom for advanced ray-traying, but a 36GBs seems so unnecessary and will just make the GPU more expensive than needed.

Literally the only value is for the coveted audience that wants to mess around with AI LLMs, 3d rendering, scientific modeling...but also play some games.

 

[Sanepar]

128GB of GDDR7 and only 4090 performance? Firstly these rumors are bullshit, secondly, you're looking at a $3000 GPU

5090 is 1.71x 4080 perf if 2.64x 4080 is true, udna flagship will be 50% faster than 5090.

 

[Perrott]

BIG NAVI

 

[Bry0]

5090 is 1.71x 4080 perf if 2.64x 4080 is true, udna flagship will be 50% faster than 5090.

If you look at the video it says that number is red because "6) I highlighted the "10090 XTX" in Red because I highly suspect the performance will be lower than the scratch math suggests if it really has only a 380W TBP."

 

[Grildon Tundy]

It'd be cool if nothing else from those specs that they do hit that VRAM.

 

[delirium trigger]

The Radeon cycle:

- wildly optimistic rumors about how much ass the next Radeon is going to kick <- WE ARE HERE

- new Radeon launches and falls short of the hype

- "yeah but just wait til we get better drivers/games are optimized for AMD because of consoles/etc"

- new GeForce launches and increases the gap

- "it's not fair to compare AMD's previous gen with Nvidia's new gen, just wait til the next Radeon launches"

- repeat



I swear this exact pattern happened with the last 4 Radeon generations at least.

My favorite is:

"Now that fsr has caught up to dlss."

Which more correctly means, the shit amd just released is about as good as this shit nvidia released last gen.

 

[SScorpio]

The Radeon cycle:

- wildly optimistic rumors about how much ass the next Radeon is going to kick <- WE ARE HERE

- new Radeon launches and falls short of the hype

- "yeah but just wait til we get better drivers/games are optimized for AMD because of consoles/etc"

- new GeForce launches and increases the gap

- "it's not fair to compare AMD's previous gen with Nvidia's new gen, just wait til the next Radeon launches"

- repeat



I swear this exact pattern happened with the last 4 Radeon generations at least.

There's a very occasional extra branch on this where NVIDIA gets scared about the rumor and releases a really good card and a fair price. The 1080ti was the last time this happened.

 

[Buggy Loop]

Every gen the stupid rumours start with MLID, like clockwork

 

[FireFly]

The Radeon cycle:

- wildly optimistic rumors about how much ass the next Radeon is going to kick <- WE ARE HERE

- new Radeon launches and falls short of the hype

- "yeah but just wait til we get better drivers/games are optimized for AMD because of consoles/etc"

- new GeForce launches and increases the gap

- "it's not fair to compare AMD's previous gen with Nvidia's new gen, just wait til the next Radeon launches"

- repeat



I swear this exact pattern happened with the last 4 Radeon generations at least.

I would say RDNA 4 was pretty much what people expected. Also there was a lot of skepticism about RDNA 2 given AMD hadn't competed in the high end for a long time. It's RDNA 3 that was the huge disappointment.

 

[Insane Metal]

And all that at 380W. That's pretty much a miracle. So yeah, that's not happening. We'll be lucky to get 1.5x 4080 perf at 400W.