Ultimate Performance, Leading Technology
H9236 is a high-performance GPU server based on Intel® Eagle Stream platform. It supports Nvidia H20 modules and is one of the best choices for AI training and large model inference.
Application
Large model training, large model inference, HPC
Features
Leading Architecture Design
Based on the 4th Gen Intel®Xeon®Scalable Processor, up to 350W TDP
Support the latest AI training modules
CPU-Switch supports dual interconnection mode, maximizing the performance of the platform
Flexible System Architecture
Flexible X16 or X32 PCIe link bandwidth between CPU and Switch
Support NVidia H20 module
Full height and half height NIC cards available
Reliable Product Quality
N+N redundant power supplies
Stable operation under high ambient temperatures with efficient heat dissipation design
Extensive experience in developing and debugging multiple modular products on the same platform
Product Specification
| Name | Taihang H9236 AI Server |
| CPU | 2x 4th Gen Intel®Xeon®Processor, up to 350W TDP |
| Memory | Up to 32x DDR5 DIMMs, 8 channels per CPU |
| Storage | 10x NVMe SSDs (optional 2x SATA)l 1x onboard M.2 SSD |
| PCIe Extension | 8x FHHL/HHHL (from Switch) 4x FHHL (from CPU) |
| Management | BMC: AST2600, 1x dedicated RJ45 MGMT port |
| IO | 2x USB, Type-A 1x PWR BTN with LED 1x UID BTN with LED 1x Status LED 1x Console PORT, Type C 1x Front VGA |
| Thermal | 15x High-performance hot-swappable fans, N+1 Redundancy |
| Temperature | 5ºC - 35ºC |
| PSU | 2x 12V PSUs for CPU module, 1+1 Redundancy 6x 54V PSUs for GPU module, 1+1 Redundancy |
| Dimension | 900mm (D) x 447mm (W) x 352mm (H) |
Series products
H8330 AI Server
AMD EPYC™ 4U8 AIC AI Server
Leading Architecture Design
Flexible Topology
Extensive Partner Ecosystem
H8230 AI Server
Intel® EGS 4U8 AIC AI Server
Leading Architecture Design
Flexible Topology
Extensive Partner Ecosystem