Brand: Intel | Category: GPUs
SKU: I-HGMD3MCMRAAAA | Part #: HGMD3MCMRAAAA | MPN: HGMD3MCMRAAAA
Contact for Pricing — Request a Quote
The Intel Gaudi 3 AI Accelerator OAM Module (HGMD3MCMRAAAA) is a passive OAM (OCP Accelerator Module) form factor accelerator designed for large-scale AI inference and training workloads in disaggregated data center architectures. Built on Intel's advanced 5nm process node, this accelerator delivers exceptional compute density and memory bandwidth to support demanding generative AI and deep learning applications.
The HGMD3MCMRAAAA features 64 Matrix Multiplication Engines (MMEs) and 8 Tensor Processor Cores (TPCs) paired with 128 GB of HBM2e memory, enabling high-throughput processing of multiple precision formats including FP8, BF16, FP16, FP32, INT8, and INT16. With 3.7 TB/s memory bandwidth and 4.8 Tb/s bidirectional on-chip network bandwidth via 24 ports of 200Gb/s RoCEv2 Ethernet, this module excels in distributed training and inference scenarios. The PCIe Gen5 x16 host interface ensures efficient system integration, while the 900W thermal design power enables flexible deployment in standard data center cooling environments. Support for PyTorch and TensorFlow through the Intel Gaudi software stack (SynapseAI SDK) simplifies framework integration, and Linux operating system support—including Ubuntu and RHEL—guarantees broad compatibility. This accelerator is engineered to the OCP OAM v2.0 specification, making it ideal for any AI infrastructure team seeking to build modular, scalable compute clusters. Contact Omnixon Global to request a quote for the Intel Gaudi 3 AI Accelerator OAM Module.
| Brand | Intel |
| Category | GPUs |
| SKU | I-HGMD3MCMRAAAA |
| Part Number | HGMD3MCMRAAAA |
| Condition | New |
| Capacity | 128 GB HBM2e and 3 |
| Manufacturer Part Number | HGMD3MCMRAAAA |
| Product Family | Intel Gaudi 3 |
| Form Factor | OAM (OCP Accelerator Module) |
| Process Node | 5nm |
| Matrix Multiplication Engines (MMEs) | 64 |
| Tensor Processor Cores (TPCs) | 8 |
| Memory Type | HBM2e |
| Memory Capacity | 128 GB |
| Memory Bandwidth | 3.7 TB/s |
| Supported Precision Formats | FP8, BF16, FP16, FP32, INT8, INT16 |
| On-chip Network Ports | 24x 200Gb/s Ethernet (RoCEv2) |
| Aggregate On-chip Network Bandwidth | 4.8 Tb/s bidirectional |
| Host Interface | PCIe Gen5 x16 |
| Thermal Design Power (TDP) | 900W |
| Supported Frameworks | PyTorch, TensorFlow (via Intel Gaudi software stack / SynapseAI SDK) |
| Operating System Support | Linux (Ubuntu, RHEL) |
| Compliance | OCP OAM v2.0 specification |
The Intel Gaudi 3 AI Accelerator OAM Module is a brand new gpus product manufactured by Intel. It has the SKU I-HGMD3MCMRAAAA and part number HGMD3MCMRAAAA. This enterprise-grade product is available from Omnixon Global.
The Intel Gaudi 3 AI Accelerator OAM Module accelerates AI/ML training, inference, scientific HPC, and virtualization (vGPU) workloads. Typical deployments include LLM training clusters, computer-vision pipelines, financial risk modeling, and rendering farms.
The key specifications of the Intel Gaudi 3 AI Accelerator OAM Module include: Max Memory: 128 GB HBM2e and 3, Performance: 7 TB/s memory bandwidth, support: 3 years standard (manufacturer), RAID Support: H100, GPU Support: NVIDIA H100, AI Optimized: Yes. Storage capacity: 128 GB HBM2e and 3. For complete specifications and technical documentation, please contact our sales team.
Key specifications for the Intel Gaudi 3 AI Accelerator OAM Module: capacity 128 GB HBM2e and 3; 3 years support; new condition; max memory 128 GB HBM2e and 3; performance 7 TB/s memory bandwidth; support 3 years standard (manufacturer); raid support H100; gpu support NVIDIA H100; ai optimized Yes. Manufacturer part number HGMD3MCMRAAAA. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
You can buy the Intel Gaudi 3 AI Accelerator OAM Module from Omnixon Global, a trusted enterprise IT hardware supplier. We serve enterprises in over 100 countries. Request a quote through our website or contact our sales team directly.