Even if the console is iterating at a regular rhythm, PC hardware has found a way to stay ahead. This is usually due to innovation that has changed the way computer graphics work, and this is now happening in Nvidia’s next wave of GeForce RTX GPUs. RTX 3070, 3080 and 3090 use Micron’s new GDDR6X memory. This is one step higher than the traditional GDDR6 VRAM, and because of Micron’s PAM4 technology, it can represent a leapfrog development in system bandwidth performance.
Pulse Amplitude Modulation (PAM4) is the name for pushing 2-bit data instead of 1-bit data through a connection. This is a technical concept, but in essence, this allows the system to be called from 4 potential states instead of 2 potential states for every information burst. Using PAM4, Micron and Nvidia are expected to significantly increase the I/O data rate and bandwidth. The power efficiency of GDDR6X is also higher. At a speed of 21
Micron claims that the per-pin data rate of GDDR6X is 21Gb/s, and when you combine all these pins and memory modules on a device like RTX 3090, the total memory bandwidth of the GPU is 936GB/s. That is huge. For comparison, the memory bandwidth of the Xbox Series X is up to 560GB/s, while the PS5 is 448GB/s.
In short, this means that the GPU can immediately extract large amounts of data from its memory. This will lead to better games.
“[Gamers] Ralf Ebert, head of Micron’s graphics department, said in a media speech. “This increases the data transferred between the GPU and the memory interface. We believe that we will achieve 8K resolution and close to real-time ray tracing.”
In practice, this means that the GPU will eventually have enough bandwidth to handle 4K comfortably. Even when the RTX 2080 Ti, which is priced at $1,200, cannot reach 60 frames per second at 4K, GDDR6X is also crucial to overcome this obstacle.