It looks like AMD’s next generation RDNA 3 ‘Navi 3X’ GPUs for Radeon RX Gaming graphics cards will have a much bigger increase in gen-over-gen performance. Following in the footsteps of its AMD Navi 33 GPU history, KittyYuko has now revealed that the flagship Navi 31 GPU will be an insanely powerful chip.
AMD Navi 31 GPU with RDNA 3 MCM design to offer 3x performance above Big Navi 21 RDNA 2 GPU, is expected to have 160 computer units and over 10,000 cores
Like the Navi 33 GPU as we reported yesterday, the AMD Navi 31 GPU will also be based on RDNA 3 (GFX11) GPU IP. The Navi 31 GPU will replace the current flagship Navi 21 GPU and will be almost twice as large given that it will have a dual-chiplet design. We all think that the current Big Navi (21) GPU is an insanely powerful chip, but based on the rumors we now hear about next generation parts, the Big Navi (31) GPU will make us forget the existing parts.
2.5x is too little.
– Yuko Yoshida (@KittyYYuko) May 2, 2021
Nothing I can confirm 100% now, but from what I know, the Navi 31 is an 80 CU chiplet and top SKU has 2 of them.
– Kepler (@ Kepler_L2) January 1, 2021
In a response to Paul Eccleston of RedGamingTech, KittyYuko said that a 2.5x performance jump to the AMD RDNA 3-based Navi 31 GPU is too little. This suggests that the next generation flagship could end up almost 3 times faster than the Navi 21 GPU. By comparison, the existing Navi 21 GPU has 80 computer devices or 5120 stream processors. The chip also has Infinity Cache and has a mold size of 520 mm2 at the 7nm process node.
The earlier rumor of the Navi 33 mainstream chip reveals that it would replace the RX 6800 and RX 6900 (RDNA 2 Big Navi 21 GPU) offerings with a similar core number, and the flagship will take things to the next level. You get double 80 Compute Unit doors, each with 5120 stream processors for a total of 10,240 stream processors. AMD RDNA 3-based Radeon RX GPU SKUs will also be based on a brand new process node from TSMC, reportedly 5 nm, which will result in space savings. Compared to the TSMC 7nm node, the N5P (5nm) delivers an improved speed of 15% or reduced power consumption by approx. 30%.
AMD has already patented an active brochiplet solution for its next-generation GPUs that comes with an integrated cache and connects the multiple dies found on Navi 3X (RDNA 3) GPUs, you can read more about that here. Other details from RedGamingTech suggested that the AMD Navi 31 GPU will have raytracing performance similar to NVIDIA’s Ampere GPUs, while the Machine Learning features for technologies like FiedilityFX Super Resolution are handled by low-precision setups rather than dedicated hardware like Tensor Cores on NVIDIA’s GPUs.
AMD RDNA 3 ‘Navi 3’ GPU lineup of Radeon RX graphics cards will compete with NVIDIA’s own MCM GPUs based on either Ada Lovelace GPUs. From now on, the next generation parts are still not expected in late 2022 or early 2023, so treat this information with a light grain of salt.