In less than two weeks, Nvidia will host its GTC keynote, where we're hoping to hear more about Ampere. But while we're waiting, we're already seeing hints of the next-gen graphics architecture in a filed trademark (opens in new tab) for the DGAX A100, as spotted by hardware leaker @Komachi Ensaka (opens in new tab).
The most noteworthy part of this name is A100. The name falls in line with the two most recent Tesla GPUs. The Pascal architecture-based GPU was the P100, with the P standing for Pascal, while the V100 used Volta. The A100 naming points to an Ampere-based Tesla GPU.
Also interesting about this trademark filing is the combination of DGX and A100 in one product name. Nvidia's DGX systems are their own custom-built systems equipped with Tesla GPUs for deep learning and AI training. Previously, however, they were simply called the DGX-1 and DGX-2. With the next system trademarked as the DGX A100, it's possible there will be different tiers of DGX systems moving forward.
So far though, Nvidia has been quite good at keeping Ampere under wraps and hasn't confirmed any information about the architecture. Nevertheless, there have been many rumors, which is why we compiled an article with everything we know about Ampere so far.
For the official announcement, tune in to Nvidia's YouTube (opens in new tab) page on May 14 at 6 a.m. PT.