multi instance gpu

A30 Enterprise GPU with ECC NVLink Video Decoder Multi-Instance GPU Product Overview The A30 is an Ampere GA100 data center GPU engineered for AI workloads, high-performance computing, and multi-tenant virtualization environments. Key Features Equipped with 24GB HBM2 ECC memory and 933GB/s high bandwidth Delivers up to 330 TOPS INT8 performance for fast AI inference and supports lightweight model training Features MIG technology, enabling division into up to 4 independent GPU

A40 Enterprise GPU The A40 is a high-end Ampere (GA102) data center GPU, purpose-built for professional visualization, ray tracing, and enterprise AI applications. Engineered for demanding workloads, it delivers exceptional performance and reliability in data center environments. Key Features Memory 48GB ECC GDDR6 with 696GB/s bandwidth Architecture Second-gen RT Cores and third-gen Tensor Cores Multi-Instance GPU Supports up to 7 MIG instances NVLink Up to 96GB combined