According to whispers from the rumor mill, AMD’s next-gen RDNA 3 based Navi 33 die is all set to be taped out (finalized) in the last quarter of 2021. Additionally, the Navi 31 (RX 7900 XT) flagship which is going to be based on an MCM (chiplet) design is allegedly going to pack a total of over 15 thousand cores (15,360 to be exact). That’s pretty insane especially if you consider that the Radeon RX 6900 XT which is based on the full-fledged Navi 21 die consists of *just* 5,120 stream processors.
We can conclude a few things from this rumor. Firstly, the Radeon RX 7900 XT (Navi 31 MCM) should easily be 2.2-2.5x faster than its predecessor, with the RX 7800 XT (Navi 31/32) beating the RX 6800 XT by 30-40% despite featuring the same core counts. A lot will depend on the process on which these chips are fabbed, as the 5nm node should allow for higher core clocks, boosting the overall performance by 15-20%. If a 7nm derivative such as N6 is used instead, then we’re probably looking at more conservative gains. It’ll also be interesting to see if AMD includes any dedicated hardware for Tensors or whether the Ray Accelerators are upgraded to account for BVH traversal.
There’s also some info on NVIDIA’s Hopper Data Center graphics cards which are also said to be based on an MCM design. According to @kopit7kimi, every GPC (Graphics Processing Cluster) in the GH100 will consist of three CPCs which in turn will feature three TPCs (Texture Processing Cluster), further sub-divided into two SMs. It’s worth noting that we are expecting a separate architecture for the gaming and data center markets, with Ada Lovelace coming to the former as a monolithic design and Hopper with its MCM design headed to the next-gen Tensor core accelerators.