Even with consoles iterating on a regular cadence, PC hardware has found ways to stay ahead. This is typically due to innovations that change how computer graphics work, and that is happening right now with Nvidia’s next wave of GeForce RTX GPUs. The RTX 3070, 3080, and 3090 use Micron’s new GDDR6X memory. This is a step above the traditional GDDR6 VRAM, and it could represent a generational leap forward in system-bandwidth performance thanks to Micron’s PAM4 tech.
Pulse amplitude modulation (PAM4) is the name for pushing 2 bits of data over a connection instead of 1 bit. It’s a technical concept, but essentially, this enables the system to call from 4 potential states instead of 2 with each burst of information. With PAM4, Micron and Nvidia are promising a significant boost to I/O data rates and bandwidth. GDDR6X is also more power efficient. At 21Gb/s, GDDR6X uses 15% less power per bit than GDDR6 at 14Gb/s.
Micron claims that GDDR6X’s per-pin data rate of 21Gb/s, when you combine all those pins and memory modules on something like the RTX 3090, the GPU has a total memory bandwidth of 936GB/s. That’s massive. For comparison, the memory bandwidth of the Xbox Series X tops out at 560GB/s and the PS5 is 448GB/s.
Put simply, this means that the GPU can pull a massive amount of data from its memory in an instant. And that’s going to lead to better games.
“[Gamers] benefit from increased system-data rates,” Micron graphics boss Ralf Ebert said in a media presentation. “That increases the data that is transferred between the GPU and the memory interface. We believe we’ll be enabling 8K resolutions and getting closer to real-time ray tracing.”
What it really means is GPUs will finally have enough bandwidth to comfortably handle