AMD Instinct MI350X GPU Accelerator 100-300000009

Brand: AMD | Category: GPUs

SKU: AMD-100-300000009 | Part #: 100-300000009 | MPN: 100-300000009

Contact for Pricing — Request a Quote

Request a Quote Contact Us

About the AMD Instinct MI350X GPU Accelerator 100-300000009

PCIe 5.0 x16 and 750 W TDP define the server chassis and power supply requirements for this AMD Instinct MI350X GPU Accelerator (part number 100-300000009), making compatibility assessment the first step in your deployment planning. This enterprise-grade accelerator is purpose-built for data center, HPC cluster, and AI infrastructure environments where large-scale compute density and memory bandwidth are mission-critical. The AMD Instinct MI350X combines cutting-edge CDNA 4 architecture with massive on-board memory resources to deliver the throughput demanded by modern AI workloads, large language models, and scientific computing applications.

Housed in the OAM (OCP Accelerator Module) form factor, this accelerator integrates seamlessly into standard data center server designs while maintaining passive cooling—requiring either liquid or forced-air chassis cooling infrastructure. With 288 GB of HBM3E memory and 8.0 TB/s memory bandwidth, the MI350X eliminates memory bottlenecks in compute-intensive workflows. AI infrastructure teams will appreciate the wide framework support: PyTorch, TensorFlow, JAX, OpenCL, and HIP all run on AMD's open-source ROCm software stack, ensuring compatibility with existing development pipelines. Peak compute performance spans FP8 (4,610 TFLOPS), FP16 (2,305 TFLOPS), BF16 (2,305 TFLOPS), FP32 (1,153 TFLOPS), and FP64 (576 TFLOPS), enabling flexible precision optimization for different model architectures. AMD Infinity Fabric interconnect enables multi-GPU scaling for distributed training and inference at scale.

Key Specifications

  • Manufacturer: AMD
  • Manufacturer Part Number: 100-300000009
  • GPU Architecture: CDNA 4
  • Memory Type & Capacity: HBM3E, 288 GB
  • Memory Bandwidth: 8.0 TB/s
  • PCIe Interface: PCIe 5.0 x16
  • Thermal Design Power: 750 W
  • Interconnect: AMD Infinity Fabric
  • Software Stack: AMD ROCm (open-source)
  • Supported Frameworks: PyTorch, TensorFlow, JAX, OpenCL, HIP

Contact Omnixon Global to request a detailed quotation for the AMD Instinct MI350X GPU Accelerator 100-300000009 and discuss your enterprise deployment requirements.

Technical Specifications

BrandAMD
CategoryGPUs
SKUAMD-100-300000009
Part Number100-300000009
ConditionNew
InterfacePCIe Gen5
Manufacturer Part Number100-300000009
Product FamilyAMD Instinct MI350X
GPU ArchitectureCDNA 4
Form FactorOAM (OCP Accelerator Module)
Memory TypeHBM3E
Memory Capacity288 GB
Memory Bandwidth8.0 TB/s
Peak FP8 Compute4,610 TFLOPS
Peak FP16 Compute2,305 TFLOPS
Peak BF16 Compute2,305 TFLOPS
Peak FP32 Compute1,153 TFLOPS
Peak FP64 Compute576 TFLOPS
TDP (Thermal Design Power)750 W
InterconnectAMD Infinity Fabric
PCIe InterfacePCIe 5.0 x16
Software StackAMD ROCm (open-source)
Supported FrameworksPyTorch, TensorFlow, JAX, OpenCL, HIP
CoolingPassive (liquid or forced-air chassis cooling required)
Target DeploymentData center, HPC cluster, AI infrastructure

Frequently Asked Questions about AMD Instinct MI350X GPU Accelerator 100-300000009

What does the AMD Instinct MI350X GPU Accelerator do?

The AMD Instinct MI350X GPU Accelerator 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.

What are the headline specs of the AMD Instinct MI350X GPU Accelerator?

Key specifications for the AMD Instinct MI350X GPU Accelerator: interface PCIe Gen5; 3 years support; new condition; support 3 years standard (manufacturer); ai optimized Yes. Manufacturer part number 100-300000009. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.

Is the AMD Instinct MI350X GPU Accelerator compatible with my infrastructure?

The AMD Instinct MI350X GPU Accelerator requires a PCIe Gen4 or Gen5 x16 slot, server power adequate for the card's TDP, and CUDA/ROCm driver support in your hypervisor or bare-metal OS. Sales engineering will confirm chassis fit (1U/2U/4U), PCIe lane count, and PSU headroom before quoting.