Why it matters: Nvidia’s GTC 2020 Keynote takes place this Thursday, where it’s almost certainly going to unveil the next-generation Ampere architecture—a “get amped” tagline leaves little doubt. To get people even more excited about the reveal, the company has uploaded a video showing CEO Jensen Huang cooking “the world’s largest graphics card.”
The 27-second video shows Huang pulling what looks like a very heavy board from the oven. It’s likely to be an Ampere-based Tesla graphics card, featuring eight GPUs. With Nvidia recently filing a trademark for DGX A100, the product is likely to be an updated version of the company’s DGX-1 deep learning system. The GPU’s, meanwhile, are suspected to be Nvidia A100, which succeed the Pascal-based P100 and Volta-based V100.
While the GTC 2020 keynote could focus on workstation products, we can expect to see the Ampere-based consumer products arrive later this year.
It seems almost certain that Nvidia will catch up with AMD by moving to 7nm with Ampere. Earlier this month, we heard that team green has put in a massive order with TSMC for its 7nm node technology. The same report claimed the firm is planning beyond Ampere and to its next-next-gen 5nm Hopper architecture, the manufacture of which could be split between TSMC and Samsung.
The second half of 2020 will be an exciting time for graphics card fans. In addition to the Ampere products, AMD will be launching its RDNA 2-based Big Navi GPUs, bringing Radeon cards that can compete with Nvidia’s high-end offerings.