Among the information announced by AMD during the Financial Analyst Day 2022 we must also report an update of our own roadmap for video cards targeting the client segment. Let’s talk about the solutions of the Radeon family, based on RDNA architecture which currently sees the RDNA 2 implementation adopted in the cards of the Radeon RX 6000 family.
The next two iterations of the desktop architecture will be called RDNA 3 e RDNA 4, with the market debut of the last one expected in 2 years. With the GPUs of the RDNA 3 family, AMD aims for a performance increase per watt of over 50% compared to what is currently offered by the proposals of the RDNA 2 family, thanks to the adoption of 5 nanometer production technology for the construction of the GPU.
The American company will also do use of new advanced packaging technologies using an a chiplet, popularized by the CPUs of the Ryzen family over the last few years but which in this case can be assumed to be of a more advanced type given the architectural peculiarities of the GPUs. Alongside this we will also find the use of the new generation of Infinity Cache, which debuted with the RDNA 2 family GPUs; they will also not fail optimization of the graphics pipeline together with a extended redesign of the Compute Units. At the moment AMD has not elaborated further these technical details, pending the official launch which should take place by the end of the current year.
Going further with the roadmap we encounter the RDNA 4 architecturefor which in addition to the name AMD has not actually provided specific indications other than the mention that its construction will be done using what is mentioned as “advanced node“. Logically, this would be something more sophisticated in terms of production technology than the 5 nanometer one adopted for the RDNA 3 GPUs, but it remains to be understood how far AMD will want and be able to go with the construction of these GPUs. In any case, there will still be a long wait, as the debut period expected for these solutions is 2024.