Huawei-FusionServer G8600 V7

Description

The Huawei FusionServer G8600 V7 is a high-end 8U GPU server engineered for computationally intensive workloads that require exceptional parallel processing capability, high-speed I/O, and large-scale accelerator deployment. Huawei positions the platform for AI training, HPC, scientific computing, image analysis, and data analytics.

The server supports two 4th and 5th Gen Intel Xeon Scalable processors, including Sapphire Rapids and Emerald Rapids platforms. Depending on the selected processor, the platform supports up to 64 cores per CPU, processor TDPs of up to 385W, and PCIe 5.0 connectivity with up to 80 PCIe lanes per processor.

For memory-intensive workloads, the G8600 V7 provides 32 DDR5 DIMM slots, supporting RDIMM or RDIMM-3DS memory and speeds of up to 5600 MT/s. With supported 256GB DIMMs, the platform can reach up to 8TB of system memory, providing substantial capacity for large AI datasets, simulations, analytics, and HPC workloads.

The primary feature of the G8600 V7 is its 8-GPU architecture. The server supports an 8-GPU NVLink module, allowing multiple GPU accelerators to operate as a tightly integrated high-performance computing platform. This architecture is particularly suited to large-scale AI model training and other workloads that benefit from high-bandwidth GPU-to-GPU communication.

The platform also provides extensive PCIe expansion. The current platform specification supports up to 13 PCIe slots, including 12 standard PCIe slots and one OCP 3.0 slot. This allows the server to accommodate high-speed networking, storage controllers, accelerators, and other expansion hardware.

Storage can be configured for either high capacity or high performance. Supported configurations include up to 8 × 2.5-inch SAS/SATA drives, up to 8 × NVMe SSDs, and 2 × M.2 SSDs. Hardware RAID options include RAID 0, 1, 10, 5, 50, 6, and 60, with optional supercapacitor protection for cached data.

The G8600 V7 incorporates a specialized high-capacity power and cooling architecture designed for GPU workloads. The GPU area uses 10 × 54V fan modules in N+1 redundancy, while the CPU area uses 5 × 12V fan modules. The platform supports multiple 3000W Titanium hot-swappable power supplies, with redundant configurations designed for high-power GPU deployments.

Enterprise management is provided by the integrated iBMC controller, which supports remote monitoring, diagnostics, configuration, hardware security, and out-of-band management. Supported interfaces include Redfish, SNMP, and IPMI 2.0, while HTML5/VNC-based remote KVM provides remote console access.

Key Specifications

Specifications vary according to the installed CPU, GPU, memory, storage, and expansion configuration.

Key Features & Benefits

• 8U flagship AI server
• Supports two 4th/5th Gen Intel Xeon Scalable processors
• Up to 64 cores per processor
• Up to 385W CPU TDP
• 32 DDR5 DIMM slots
• Up to 5600 MT/s memory speed
• Up to 8TB system memory with supported DIMMs
• 8-GPU accelerator architecture
• NVLink-supported GPU module
• PCIe 5.0 expansion
• Up to 13 PCIe expansion slots
• OCP 3.0 networking
• High-speed NVMe storage
• Hardware RAID support
• High-capacity Titanium power supplies
• N+1 GPU cooling redundancy
• Hot-swappable power supplies and drives
• Integrated iBMC remote management
• Designed specifically for large-scale accelerated computing

Recommended Applications

The Huawei FusionServer G8600 V7 is suitable for:

• AI Model Training
• Deep Learning
• Large Language Model (LLM) Training
• High-Performance Computing (HPC)
• Scientific Computing
• AI Inference
• Image Analysis
• Video Analytics
• Big Data Analytics
• Computer Vision
• Engineering Simulation
• Research and Development
• GPU-Accelerated Computing
• Enterprise AI Infrastructure