Nvidia will use GDDR6X memory in at least the RTX 3090, if not other high-end Ampere graphics cards, according to a technical overview released by memory maker Micron. This news comes from a PDF from Micron (discovered by Videocardz), which clearly states that the next generation graphics card will feature the as yet unannounced GDDR6X, rated at 19-21Gbps.

This is very fast memory, and according to the same document [PDF alert] on Micron's website, the company expects GDDR6X cards to average about 12GB. Both are increasingly important indicators for high-resolution and high-fidelity gaming; if 4K60 is coveted, it is not simply a matter of more cores.

For comparison, the RTX 2080 Ti Founders Edition was equipped with 11GB memory rated at 14Gbps.

This is also the first official information we received about the existence of GDDR6X memory configurations. Nvidia has made a similar move with its high-end Pascal graphics cards with shiny new GDDR5X memory. speeds considerably faster than the GDDR5X, increasing from 10 Gbps to 7 Gbps and then to 11 Gbps, respectively. This means that GDDR6X may run at slightly lower voltages (and may also be more expensive). [It is unlikely that GDDR6X will become widespread across the GPU stack.

This document also provides further evidence of Nvidia's choice of branding and naming schema for the RTX 30 series, as well as the highly touted name RTX 3090. However, since Micron does not assemble the cards, only sells the chips, some guesswork may be required on Micron's part. Looking again at the included table, it seems that the exact RTX 3090 graphics card specifications cannot be determined from what is listed.

It is likely that we are hearing a well-established name for the GPU generation, as Nvidia is expected to announce the Ampere generation at the GeForce Special Event on September 1, 2020. The proximity of this event also adds weight to the credibility of the recent spec controversy at Micron.