OCI GPU Quick Start: NVIDIA H200

This document provides hardware specifications, supported OS images, onboarding verification, sample benchmarks, and best-practices for OCI deployments using the NVIDIA H200 GPU shape.

BM.GPU.H200.8 is a high-bandwidth NVIDIA H200 bare-metal shape intended for large-scale AI and HPC workloads, with eight 141 GB GPUs, dual Intel Xeon Platinum 8480+ processors, 3 TB of DDR5 memory, and RoCE-capable scale-out networking.

At a Glance

• Shape: BM.GPU.H200.8
• GPU configuration: 8 x NVIDIA H200 141 GB
• Recommended OS baseline: Oracle Linux 8 or 9, Ubuntu Linux 22.04, or Ubuntu Linux 24.04
• Recommended software baseline: DOCA OFED 3.2.1, NVIDIA Driver 580/590 (Open), CUDA 13.0/13.1
• Primary verification command: nvidia-smi
• Operational profile: scale-out AI and HPC with single-subnet NCCL guidance

Table of Contents

• OCI GPU Quick Start: NVIDIA H200
  • At a Glance
• Table of Contents
• Hardware Specifications
• Recommended Operating Systems
  • Recommended Software Version
  • Custom OS Image Creation with Packer
  • Provided Images
  • Hello World Verification
• Performance Benchmarks
  • All Reduce - Single Node
  • Multi-node Guidance
  • Model Inference Performance
• OKE GPU Getting Started
• Troubleshooting
  • GPU Visibility
  • NUMA Layout
  • RDMA Interface State
  • DCGM Diagnostics
  • PCIe and NVLink Validation
• Further Reading \& Support

Hardware Specifications

See the OCI Compute Shapes Docs for up-to-date details.

Recommended Operating Systems

• Oracle Linux 8
• Oracle Linux 9
• Ubuntu Linux 22.04
• Ubuntu Linux 24.04

Recommended Software Version

• DOCA OFED 3.2.1
• NVIDIA Driver 580 or 590 (Open)
• CUDA 13.0 or 13.1
• Oracle Cloud Agent 1.57.0
• Use the Provided Images table below for the current validated OCI image combinations

Custom OS Image Creation with Packer

To build your images using packer clone the OCI HPC Images repo and run the commands found there OCI HPC Images GitHub Repo.

Provided Images

Hello World Verification

Run nvidia-smi to verify that all eight GPUs are visible and healthy:

nvidia-smi

You should see all eight H200 GPUs listed with a healthy driver stack and expected memory footprint.

Performance Benchmarks

NVIDIA publishes NCCL as the primary collective communication library for multi-GPU AI and HPC workloads. The source material for H200 focuses on representative single-node and multi-node NCCL workflows plus the network guidance required for scale-out runs.

• All Reduce - Single Node
• Multi-node Guidance
• Model Inference Performance

All Reduce - Single Node

./build/all_reduce_perf -b 8 -e 8G -f 2 -g 8

Multi-node Guidance

The source material highlights a single-subnet network layout for best multi-node NCCL behavior on this family.

For guidance on running additional NCCL collective benchmarks on this GPU family, see the NCCL user guide.

Model Inference Performance

The H200 source page in this workflow is oriented more toward validation, NCCL guidance, and training examples than a clean standalone inference table, so no normalized model-performance table is included here.

OKE GPU Getting Started

Information on getting up and running on OKE can be found here.

Troubleshooting

This guide includes a broad health-check set covering GPU visibility, NUMA topology, RDMA connectivity, DCGM diagnostics, PCIe bandwidth, and NVLink validation.

• GPU Visibility
• NUMA Layout
• RDMA Interface State
• DCGM Diagnostics
• PCIe and NVLink Validation

GPU Visibility

nvidia-smi

NUMA Layout

numactl --hardware

For H200, the source material expects a layout comparable to a 112-core dual-socket system, or fewer visible cores when hyperthreading is disabled.

RDMA Interface State

rdma link

The source material identifies the front-end network as eth0 on mlx5_1, with RoCE RDMA interfaces expected to report ACTIVE and LINK_UP.

DCGM Diagnostics

dcgmi diag -r 1
dcgmi diag -r 2
dcgmi diag -r 3

The source material describes:

• r1 as a basic metadata and deployment check
• r2 as a deeper integration and hardware check
• r3 as a fuller stress-oriented validation pass

PCIe and NVLink Validation

Additional validation tools referenced in the source material:

• bandwidthTest from NVIDIA cuda-samples for PCIe bandwidth
• nvbandwidth for NVLink bandwidth validation

Further Reading & Support

Additional references:

• NCCL User Guide
• NVIDIA nvbandwidth