NVIDIA H100 PCIe 80GB HBM2e GPU — Dell PowerEdge R750xa / R760 Add-in Card

The NVIDIA H100 PCIe 80GB HBM2e GPU is a flagship accelerator built on NVIDIA's Hopper architecture, engineered for demanding enterprise AI, high-performance computing, and data analytics workloads. Featuring the GH100 GPU die manufactured on TSMC's 4N process, the H100 PCIe variant delivers 51 TFLOPS of FP64 Tensor Core performance and 3,958 TFLOPS of FP8 Tensor Core performance, supported by 80 GB of HBM2e memory with 2 TB/s of memory bandwidth. The card connects to the host system via a PCIe Gen 5 x16 interface, enabling broad compatibility within supported server platforms without requiring NVLink fabric topology.

Configured as a Dell-qualified add-in card for the PowerEdge R750xa and R760 server platforms, this GPU carries Dell's component validation and platform integration for enterprise deployments. The H100 PCIe incorporates NVIDIA's fourth-generation Tensor Cores with support for FP8, FP16, BF16, TF32, FP64, and INT8 data types, alongside the Transformer Engine — purpose-built for large language model (LLM) training and inference workloads. Confidential Computing support via hardware-based TEE (Trusted Execution Environment) enables secure multi-tenant AI deployments in regulated environments.

With a 350W TDP and passive cooling design suited to the airflow characteristics of rack-mounted enterprise servers, the H100 PCIe 80GB is positioned for datacenter installation within standard rack infrastructure. NVIDIA's MIG (Multi-Instance GPU) technology allows the card to be partitioned into up to seven isolated GPU instances, enabling concurrent, quality-of-service-governed workload consolidation across multiple tenants or workload types. The card supports NVIDIA's full CUDA ecosystem, cuDNN, TensorRT, and the broader NVIDIA AI Enterprise software stack, making it a production-ready foundation for AI model development, scientific simulation, and data center scale inference.

Ideal for

• Large language model (LLM) training and fine-tuning for enterprise AI development teams requiring high-throughput FP8 and BF16 Tensor Core compute
• Real-time and batch AI inference serving for natural language processing, computer vision, and recommendation system models in production datacenter environments
• High-performance computing (HPC) simulation workloads in scientific research, computational fluid dynamics, and molecular dynamics requiring FP64 double-precision throughput
• Secure multi-tenant AI workload consolidation using NVIDIA MIG partitioning within regulated industries such as financial services and healthcare
• Data analytics acceleration and GPU-accelerated database query processing at enterprise scale using RAPIDS and cuDF frameworks
• AI model development and MLOps pipeline acceleration within Dell PowerEdge R750xa or R760-based on-premises AI infrastructure

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.