AMD's next-gen RDNA 3 GPU architecture is really going to open up a can of whoop-ass on NVIDIA in 2022, with the MCM-based (multi-chip module) graphics card now rumored with new 3D Infinity Cache technology.
In a new tweet, leaker Greymon55 simply teases "3D IFC" which would imply "3D Infinity Cache". The next-gen RDNA 3 architecture from AMD is expected to have multiple GPU chiplets (MCM technology, or multi-chip module) and should -- at least from this new rumor -- split the Infinity Cache across the GPU chiplets in its MCM design, to 256MB per chip.
AMD's next-gen Navi 31 GPU is expected to have 512MB of Infinity Cache, but now this could be split into a next-gen 3D stacked design through 3D Infinity Cache technology. 3D IFC hasn't been formally announced just yet, so this is rumor territory as usual. AMD is using its 3D V-Cache technology on its next-gen EPYC Milan-X processors, with a huge 768MB of L3 cache expected on the flagship EPYC 7773X (64 cores, 128 threads) processor.
- Read more: GeForce RTX 4090 rumor: 2x RTX 3090 perf, power: 'can you accept it?'
- Read more: AMD's next-gen Navi 31 GPU taped out, flagship RDNA 3 silicon is close
- Read more: AMD's next-gen Navi 31 GPU for next-gen Radeon RX 7900 XT teased again
- Read more: AMD's next-gen Navi 31, Navi 32, Navi 33 GPU block diagrams teased
- Read more: NVIDIA's next-gen AD102 + AMD's next-gen Navi 31 use 400-500W power?!
- Read more: AMD's next-gen Radeon RX 7900 XT: new Navi 31 GPU, battles RTX 4090