It looks like the next generation of AMD RDNA 3 ‘Navi 3X’ GPUs for Radeon RX Gaming graphics cards will have a much bigger performance boost than the general generation. Following its story for the AMD Navi 33 GPU, KittyYuko has already revealed that the flagship Navi 31 GPU will be an incredibly powerful chip.
AMD Navi 31 GPU with RDNA 3 MCM design to offer 3x performance improvement over the large Navi 21 RDNA 2 graphics processor, expected to include 160 units and over 10,000 cores
Similar to the Navi 33 GPU we announced yesterday, the AMD Navi 31 GPU will also be based on the RDNA 3 (GFX11) GPU IP. The Navi 31 GPU will replace the current flagship Navi 21 GPU and will be almost twice as big, given that it will feature a dual-chip design. We all think that the current Big Navi graphics processor (21) is an insanely powerful chip, but based on the rumors we now hear about the next generation of parts, the Big Navi graphics processor (31) will make us forget the existing parts.
2.5x is too little.
– Yuko Yoshida (@KittyYYuko) May 2, 2021
I can’t confirm anything 100% now, but from what I know, the Navi 31 is an 80 CU chip and the best SKU has 2 of them.
– Kepler (@ Kepler_L2) January 1, 2021
In response to Paul Eccleston of RedGamingTech, KittyYuko said the 2.5-fold jump in performance for the Navi 31 GPU based on AMD RDNA 3 was too small. This suggests that the next-generation flagship could be almost 3 times faster than the Navi 21 GPU. By comparison, the existing Navi 21 GPU has 80 units of account or 5,120 stream processors. The chip also has Infinity Cache and has a matrix size of 520mm2 per 7nm process node.
The previous rumor about the main Navi 33 chip reveals that it will replace the RX 6800 and RX 6900 (RDNA 2 Big Navi 21 GPU) with a similar number of cores and the flagship will take things to the next level. You will get double 80 matrices for computing modules, each with 5120 stream processors for a total of 10 240 processors. Radeon RX’s AMD RDNA 3 GPU SKUs will also be based on a brand new process node from TSMC, according to 5nm, which will save space. Compared to the TSMC 7nm node, the N5P (5nm) provides a 15% improved speed or about a 30% reduction in power consumption.
AMD has already patented an active bridge chipset solution for its next-generation GPUs, which comes with an integrated cache and connects the many matrices presented on the Navi 3X (RDNA 3) GPUs, you can read more about this here. Other details from RedGamingTech suggest that the AMD Navi 31 GPU will include beam tracking performance similar to NVIDIA Ampere GPUs, while machine learning capabilities for technologies such as FiedilityFX Super Resolution will be handled by low-precision operating systems rather than from special hardware like Tensor Cores to NVIDIA GPUs.
The AMD RDNA 3 ‘Navi 3’ GPU line of Radeon RX graphics cards will compete with NVIDIA’s own MCM graphics processors based on Ada Lovelace graphics processors. At the moment, parts of the next generation are not yet expected until the end of 2022 or the beginning of 2023, so treat this information with a little grain.