Intel Gaudi 3 OAM 96GB HBM2E Mezzanine Module — Dell PowerEdge XE9680 (Gaudi Config)
Brand: Dell | Category: GPUs
SKU: HLS-GAUDI3-OAM-DELL | Part #: HLS-GAUDI3-OAM-DELL | MPN: HLS-GAUDI3-OAM-DELL
Contact for Pricing — Request a Quote
About the Intel Gaudi 3 OAM 96GB HBM2E Mezzanine Module — Dell PowerEdge XE9680 (Gaudi Config)
The Intel Gaudi 3 OAM 96GB HBM2E Mezzanine Module is a passive OCP Accelerator Module (OAM) designed for seamless integration into the Dell PowerEdge XE9680 (Gaudi 3 OAM Configuration). This accelerator module delivers 96GB of HBM2E memory paired with 64 Tensor Processor Cores (TPCs) and 8 Matrix Multiplication Engines (MMEs), enabling high-throughput AI training and inference workloads at scale. The module supports multiple precision formats—FP8, BF16, FP16, INT8, and FP32—ensuring flexibility across diverse machine learning frameworks and model architectures.
The module integrates native 21-port 200GbE RoCE (RDMA over Converged Ethernet) on-board networking and a host-independent 200GbE scale-out fabric for inter-accelerator connectivity, allowing up to 8 OAM modules per Dell PowerEdge XE9680 chassis. Cooling is managed through the chassis-integrated thermal system, and management is unified via BMC and iDRAC9 integration. The Intel Habana SynapseAI SDK provides comprehensive software support, with native framework integration for PyTorch and TensorFlow. For AI infrastructure teams planning large-scale distributed training clusters, this architecture offers enterprise-grade reliability and performance density. Part number: HLS-GAUDI3-OAM-DELL.
Enterprise Deployment Scenarios
- Multi-accelerator training clusters leveraging up to 8 modules per chassis for distributed deep learning pipelines
- Large language model (LLM) fine-tuning and inference at scale using native PyTorch and TensorFlow integration
- High-precision and low-precision (FP8/INT8) model optimization across heterogeneous workload portfolios
- RDMA-accelerated inter-node communication for tightly-coupled distributed AI workflows in data centers
- Rack-density optimization for cost-efficient AI cluster deployments in enterprise environments
Contact Omnixon Global to request a detailed quotation and technical consultation for your accelerator requirements.
Technical Specifications
| Brand | Dell |
| Category | GPUs |
| SKU | HLS-GAUDI3-OAM-DELL |
| Part Number | HLS-GAUDI3-OAM-DELL |
| Condition | New |
| Manufacturer Part Number | HLS-GAUDI3-OAM-DELL |
| Accelerator Architecture | Intel Gaudi 3 |
| Form Factor | OAM (OCP Accelerator Module) Mezzanine |
| Compatible Platform | Dell PowerEdge XE9680 (Gaudi 3 OAM Configuration) |
| HBM Capacity | 96GB HBM2E |
| Tensor Processor Cores (TPCs) | 64 |
| Matrix Multiplication Engines (MMEs) | 8 |
| Supported Precisions | FP8, BF16, FP16, INT8, FP32 |
| On-Board Networking | 21-port 200GbE RoCE (RDMA over Converged Ethernet) |
| Inter-Accelerator Connectivity | Integrated 200GbE scale-out fabric (host-independent) |
| Maximum Accelerators per Chassis | 8 OAM modules (Dell PowerEdge XE9680) |
| Software Ecosystem | Intel Habana SynapseAI SDK |
| Framework Support | PyTorch, TensorFlow (via SynapseAI integration) |
| Thermal Design | Chassis-integrated cooling via Dell PowerEdge XE9680 thermal system |
| Management Interface | BMC / iDRAC9 integration via host chassis |
| Product Line | Intel Gaudi 3 AI Accelerator |
Frequently Asked Questions about Intel Gaudi 3 OAM 96GB HBM2E Mezzanine Module — Dell PowerEdge XE9680 (Gaudi Config)
What server platforms accept the Intel Gaudi 3 OAM 96GB HBM2E Mezzanine Module — Dell PowerEdge XE9680 (Gaudi Config)?
Reference servers include Dell PowerEdge XE9680 / XE9712, HPE Cray XD670, Lenovo ThinkSystem SR685a / SR675 V3, Supermicro AS-A21GE / SYS-821GE, Gigabyte G593 / G894, ASUS ESC. Share your target platform in the RFQ and we will confirm chassis-to-GPU compatibility and recommended NIC pairing.
How long is the lead time on AI GPUs?
Highly model-dependent. L40S / RTX-class: typically 3-6 weeks. H100/H200/B200 in SXM form factor: 12-16 weeks for whole-platform allocations. We quote genuine-channel ETAs only — no grey-market promises.
Do you supply matched networking (Quantum InfiniBand / Spectrum-X)?
Yes — Omnixon stocks the full NVIDIA networking lineup (Quantum-2 / Quantum-X InfiniBand, Spectrum-X Ethernet, ConnectX NICs, BlueField DPUs) so we can quote a complete training-cluster BOM, not just the GPUs.
Can you help with NVIDIA AI Enterprise licensing?
Yes. We hold genuine-channels for NVIDIA AI Enterprise software subscriptions. Add it to your RFQ and we quote node-aligned licensing along with the hardware.