NVIDIA GeForce RTX 4080 Rumors Point To Higher Memory Bandwidth And Lower Power
NVIDIA Ada Lovelace GPU rumors continue to swirl, as we are still perhaps two months away from the official launch of the first products. Earlier today, an NVIDIA GPU leaker of some repute shared what he referred to as “some updates” on the GeForce RTX 4080 model. The key purported changes are that the memory chips appear to have been upgraded to 23 Gbps, and that the total card power is now a much more modest 340W.
It is easy to imagine NVIDIA engineers and marketers fervently finalizing the specs of the RTX 40 series of graphics cards, with so many factors to consider in the current landscape. In addition to considering the details of its own products, NVIDIA must also strategize in light of competing offerings that are coming down the pipeline. And all of this is happening at a time of great difficulty in the semiconductor industry, so it's no surprise that some rumored specs remain in flux.
The first change highlighted in this latest leak is that the RTX 4080's memory spec has been upgraded from 21 Gbps GDDR6X memory to 23 Gbps. It is odd to call this an upgrade because this is still a rumor after all, but if what has been Tweeted is true the newer memory configuration is nearly 10% faster than what was previously reported.
Another quite significant change comes to the power requirements for the RTX 4080 graphics card. According to the new data the “total board power” (TBP) has been set at 340W, for reference models at least. Before today, the TBP of the RTX 4080 was rumored to be nearly 25% higher, at 420W. If the power reduction is genuine, NVIDIA's engineers have done some worthwhile optimizing, or perhaps decided it is prudent to wind back the voltage / clocks to compete with what they reckon AMD is going to do with the Radeon 7000 family.
Other key specs from Kopite7Kimi appear to have remained the same, from the GPU code name, to CUDA core count of 9,728, to the 16GB of memory connected via a 256-bit bus.
