AMD Networking Solutions

AMD Pensando DPUs and AI NICs move networking, security and storage services onto programmable silicon in the server, and keep GPU clusters running at full speed. IIS Technology designs and integrates AMD networking into compute and AI platforms so the fabric is never the bottleneck.

Why AMD networking

· Free up CPU cores. Offload virtual networking, firewalling, encryption and storage protocols to the DPU and return host cores to applications.

· Zero-trust at every server. Stateful firewalling and micro-segmentation enforced in hardware, at line rate, on every host.

· AI fabrics that stay fast. Intelligent load balancing, fast failover and selective retransmission keep GPU-to-GPU traffic moving during congestion.

· Open standards. Ultra Ethernet Consortium (UEC) and UALink support means no proprietary fabric lock-in.

· Fully programmable. P4-programmable pipelines adapt to new protocols and services through software.

Where it fits

· AI and GPU clusters — back-end scale-out fabrics for AMD Instinct and other accelerators.

· Private cloud and virtualization — offloaded SDN and security for VMware, OpenShift and Kubernetes.

· Regulated industries — hardware-enforced segmentation and encryption for compliance.

· Storage-intensive environments — NVMe-oF acceleration and data reduction.