Intel Xe-HPG architecture GPU will also be used in workstations and data centers
Previously, Raja Koduri, Intel’s senior vice president, chief architect, and general manager of the architecture, graphics, and software division, said that he will focus on Xe-HPG (for games) and Xe-HPC (for high-performance computing) architectures, both series of products (Alchemist graphics card and Ponte Vecchio) will be commercially produced, but the Xe-HP architecture GPU does not have this plan and is only used for its own cloud server.
The Xe-HP architecture was originally oriented to the cloud and media transcoding and decoding fields. Previously, Arctic Sound series computing cards equipped with GPUs of this architecture were exposed. Considering that Intel spent a lot of time and energy promoting this series of products, it is a bit surprising to finally make such a decision. Recently Raja Koduri explained this problem again, saying that Xe-HP architecture GPU is to build a developer ecosystem for HPC, AI, and visual cloud, and Xe-HPG architecture GPU will replace Xe-HP architecture on the product roadmap.
The visual cloud includes media processing, media analysis, immersive media, cloud graphics processing, and cloud gaming, which means that Xe-HPG architecture GPUs can be used for services such as GeForce NOW or Google Stadia. In addition, it can also be deployed to a media transcoding server for TV broadcasting. At the Tokyo Olympics this year, Intel, as a sponsor, provided servers built on Xeon processors to provide services for 8K, 60fps, and HDR video broadcasting. Xe-HPG architecture GPUs can further expand Intel’s influence in this field.
Xe-HPG architecture GPU will also support professional software such as 3DS Max, which means that it will appear in the workstation market, and its role should be similar to Nvidia’s Ampere architecture and AMD’s RDNA 2 architecture. The application coverage is larger than originally expected.
