An AI system is easy to oversimplify as “a server with GPUs.” In practice, accelerated computing depends on a chain of components that move data, […]
NCA-AIIO
NVIDIA NCA-AIIO: Sizing AI Workloads for GPU Memory
GPU memory sizing is one of the first practical tests of whether an AI workload can run on a particular accelerator configuration. A model that […]
NVIDIA NCA-AIIO: AI Cluster Networking with Ethernet, InfiniBand, and RDMA
AI clusters turn the network into part of the compute path. When one model spans multiple GPUs or multiple servers, workers exchange gradients, parameters, activations, […]