Now that Nvidia has finally pulled back the curtain on consumer Ampere cards, Micron has come out and officially announced its new GDDR6X memory (previously leaked) as well, providing technical details on why they care. In short, it is a much faster memory standard than GDDR6. [This is a multi-signaling technology that has been around for several years and is used in areas such as high-end networking (think 200 or 400 Gigabit Ethernet solutions). [Our multi-level signaling innovation in GDDR6X breaks the limits of traditional bandwidth and delivers record speeds," said Gary H. Harris, president and CEO of GDDR6X. Unlike conventional memory, GDDR6X features unparalleled data transfer rates, allowing us to innovate for gaming and data-intensive applications."

The adoption of PAM4 effectively doubles the number of signal states on the GDDR6X memory bus. Whereas conventional GDDR6 memory relies on two signal levels to transmit data as 1s or 0s and has a graphics memory limit of 64 GB/s, PAM4 uses four different levels to transmit two bits of data to and from memory simultaneously. [As a result, Micron's GDDR6X dramatically increases the memory bandwidth of each component to 84GB/second and enables system bandwidths of up to 1TB/second, once thought impossible," Micron explained.

Two of Nvidia's newly announced Ampere cards have moved to Micron's GDDR6X memory, priced at $1,499 for the GeForce RTX 3090 and $699 for the GeForce RTX 3080. $499 GeForce RTX 3070 still uses GDDR6.

At the moment, Micron is mass producing GDDR6X at 8 gigabit (Gb) density, with speeds ranging from 19-21 Gb/sec. However, next year Micron plans to double the density to 16Gb, possibly paving the way for a "Ti" or "Super" upgrade of Nvidia's latest generation GeForce RTX cards (this is pure speculation).