RDNA4 will be N4-based, targeting to launch in H2 2024. We could probably see CDNA4 and RDNA4 launching at the same year, and we’ll get some inspirations about what RDNA5 will look like from CDNA4’s packaging.

I hope they eventually implement a proper dedicated unit like Nvidia and Intel have.

That’s what RDNA4 will introduce.

Bought my RVII in 2019 for $699 + tax and sold it for $2000 in 2021. Perfect card for mining.

I don’t a die shrink from N7 to N6 brings any improvement of power effiency. It’s just a cost reduction because of higher density.

They can still packed the GCN driver with their RDNA driver and tell you the products are still supported, but you know it’s just a psychological comfort.

Did your source ever mention there is probably a dual-chip design of RDNA4? Some people hinted that for me after the rumor which said high-end RDNA4 chips got canceled came out.

If AMD not making a chip that is on par or bigger than nvidia’s flagship then the whole MCM design becomes meaningless. They are getting penalty from PPA and power effiency due to the interconnection between chips. What I expected is that AMD pushes the performance of their flagship to the further level, but what I see is that they are betting on higher frequency with a much smaller shader array (very ATI-style) while using MCM to save cost. Fortunately they failed, but they will never learn.

From the images that this article quotes, from Q2 2022 to Q3 2022 AMD’s dGPU market share vaporizes by its half. How did they get that data? Btw Nvidia launched their Ada GPUs on Q4 2022, which shouldn’t be considered as a factor.