Wccftech reports that Micron is aiming for 36Gbps of bandwidth per pin, while the current maximum is around 22Gbps in Nvidia’s GDDR6X solutions, and 20Gbps for AMD’s GDDR6 options. Upgrading the bandwidth per pin will drastically boost the memory bandwidth for each GPU equipped with GDDR7. For instance, a budget card with a 128-bit bus will now offer 576Gbps of bandwidth, which is a huge step up. High-end models, like the RTX 4090, will be able to hit a massive 1.7TB/s in memory bandwidth versus the current maximum of 1TB/s.
Comments
(._.)
There's nothing here…