According to new rumors, AMD may be planning to upgrade its Navi 2X GPUs with the new 6nm manufacturing node to fill a gap in the mid-range to an entry-level segment of its next generation of graphics cards, the Radeon RX 7000. What does this mean? Essentially, not all RX 7000s will have RDNA 3 architecture, but there would be RDNA 2 models with the new 6nm lithography or, put another way, AMD will go back to using “rehashes” in a new generation of graphics cards.
It is not the first time AMD has used the technique of “rehashes” in its graphics card portfolio, and it is something that users have grown accustomed to with this business in the past. For those unfamiliar with the word, it refers to accomplishing exactly what we just described: introducing an old architecture “enhanced” in a new generation of GPU that debuts a different architecture.
Only AMD’s “top” RX 7000 GPUs will be RDNA 3
According to the source of this claim, AMD’s next Radeon RX 7000 series may have both RDNA 2 and RDNA 3 architectural GPUs (very reliable if history is any guide). For the next generation (Navi 31 and 32), the company may deliver new 5 nm GPUs based on the new RDNA 3 architecture. Navi 32) while revitalizing the present RDNA 2 design in the lower manufacturer segments.
However, this does not appear to be a simple rebranding. Some existing Navi 2X series chips would receive an “optical shrink” at the 6 nm node in an attempt to improve their performance per watt consumed power ratio. Some of the performance/power consumption improvements could, of course, be used to increase operating speeds and, thereby, performance (in this case, while maintaining power consumption).
Among these GPUs are the Navi 22, which features 40 RDNA 2 compute units and a 192-bit GDDR6 memory bus, and the Navi 23 features 32 RDNA 2 compute units and a 128-bit GDDR6 memory bus. The updated Navi 22 processor will power the SKU that will succeed the current RX 6600 XT, while the upgraded Navi 23 processor will power the entry-level RX X500 series.
If this is the case, AMD will almost certainly take advantage of the opportunity to improve the video and acceleration components of the graphics, as it did with the most recent Vega series revisions. Meanwhile, RDNA 3 will most likely only power the generation’s two largest chips, the Navi 31 and Navi 32, which are 5nm and will be at the heart of the models that will supplant the firm’s existing Radeon RX 6900/6800.
AMD is back with “rehashes,” but are they out of ideas?
We’ve already highlighted that this isn’t the first time AMD has used rehashes to fill models in its graphics card generations, and it’s unlikely to be the last. It is a perfectly valid technique as long as they correctly indicate which architecture equips each graphics card (after all, they are simply launching two different architectures and two different lithographies within the new GPU generation rather than all of it is based on a new architecture as expected). Still, it indicates some weakness in the ranks of those in red.
After all, when you launch a new generation of graphics cards, users expect an increase in performance and all the goodness that a completely new graphics architecture gives. Suppose AMD’s next generation’s high-end models incorporate RDNA 3 architecture (which will, of course, be significantly more expensive than the other models). In that case, this can be extremely frustrating since it limits the new architecture’s availability to many consumers.
This technique of rehashes also denotes, as we were saying, some weakness in the brand’s engineering group. As we were saying, the idea is to launch a new generation with a new architecture covering entry-level to high-end models, and not doing so may mean that they have not been able to do it, either by development, cost, or time.