Aimed at high-data-rate AI networks, the AMD Pensando Pollara 400 network interface card is the first NIC compliant with the Ultra Ethernet Consortium 1.0 specification.| Network World
The MI355X accelerator delivers up to 40% higher token-per-dollar efficiency compared to rivals, AMD claims.| Network World
Brium addresses a key gap in enterprise AI deployment — reliance on CUDA-optimized toolchains.| Network World
This could enhance AMD’s ability to develop high-performance systems, bolstering its competitiveness against Nvidia.| Network World