AMD launches new DPUs to compete with Nvidia in the field of artificial intelligence and networking.
The new AMD Pensando DPU Salina aims to drive the next generation of infrastructure for artificial intelligence.
AMD has unveiled an enhanced data processing unit (DPU), aiming to position itself as a key player in the next generation of artificial intelligence. The new DPU, called Pensando Salina, represents the third generation of this technology and promises to double performance, bandwidth, and scalability compared to its predecessor.
The company claims that this new processor can support 400G performance, which means faster data transfer rates than ever, a significant advantage for companies seeking faster and more efficient infrastructures that align with the demands of AI.
Like its previous versions, AMD's latest DPU is divided into two components: a front-end that provides data and information to an AI cluster, and a back-end that manages the data transfer between accelerators and clusters. Along with the Pensando Salina DPU, the company has also announced the AMD Pensando Pollara 400, which will handle the back-end.
This latest device, known as the first AI-ready NIC for the Ultra Ethernet Consortium (UEC), is compatible with next-generation RDMA software and features an open networking ecosystem, giving customers the flexibility needed to adapt to the new age of AI.
Both the AMD Pensando Salina DPU and the AMD Pensando Pollara 400 are already in testing with select customers, with a public launch scheduled for the first half of 2025.
