AMD Radeon RX 7000 5 and 6 nanometers

The new mercy a somewhat indiscreet update of its profile on LinkedIn (which was corrected) from Brian Walters, AMD engineer for over 27 years, some of the most interesting data about the company's next generation of graphics adapters.

And it so happens that in such a job description, he mentioned several of the company's projects in which he participated, data that have allowed him to obtain interesting conclusions about future projects. AMD Radeon RX 7000.

The first thing is that, depending on the model, he chip It will be able to be developed in five or six nanometers.

Specifically, it appears that 5nm lithography will be reserved for high-end models, i.e. the Navi 31 and Navi 32 chips, although it is also possible that some of these will be based on 6nm, and that would be AMD's entry-level version. Radeon RX 7000, Navi 33, which would be on the six nanometer scale.

Another essential note is that the AMD Radeon RX 7000 series I would only maintain the monolithic design in the entry models, that is, those based on Navi 33, while in the rest the technological Santa Clara would make the leap to an MCM (Multi Chip Module) design.

This is a long-awaited leap, which we have been talking about for several years now, and which we learned a little over a year ago that it would debut, in terms of adapters brand graphics, via RDNA3.

Let us keep in mind that the MCM design, which It has now been present for several generations in the central units of prosecution from AMD, allows the use of multiple small chips that, when complemented in the same package, provide enormous performance.

This is, in fact, one of the reasons for the development in performance of the successive generations of the Zen architecture, and one of the considerable hopes for the performance of the AMD Radeon RX 7000 series, which thanks to this type of architecture can offer a very important leap in terms of its performance.

By integration, we hope that Navi 31 (the maximum of range) is responsible for powering the Radeon RX 7900 and 7800 series.

I could have a whopping 15,360 cores and 240 balance units.

It will have a 256-bit data pool and will mount memory GDDR6 and 512 megabytes of Infinite Cache.

Its TBP would be around 400 watts.

It could have five and six nanometer editions and, in both cases, it would use the MCM architecture.

Navi 32 would be the engine of the Radeon RX 7700 series, with 10.240 cores and 160 balance entities, 192-bit memory bus and 384 megabytes of Infinite Cache. In the same way as in Navi 33, we would find integrated MCMs on the five and six nanometer scales.

And as regards Navi 33, in this case we are talking about one that is produced in six nanometers with a monolithic design, with 5,120 cores, 40 balance entities, bus of memory 128 bits and 256 megabytes of cache.