The GPX XT6-24S3-8NVLINK GPU-accelerated server is a 4U, 8-GPU system designed
to support highly parallel workloads. Users can expect strong performance from
this system in HPC and AI applications.
With support for 8 HGX A100 SXM4 GPUs interconnected with NVIDIA’s
proprietary GPU-to-GPU Interconnect NVLINK, 3rd Gen Intel® Xeon® Scalable
Processors, 4TB of DRAM, and ten hot-swappable NVMe/SAS/SATA drives, the GPX
XT6-24S3-8NVLINK is the ideal high-performance GPU-accelerated server for highly
parallel HPC and AI workloads.
Ideal GPU Optimized Server Use Cases
AI / Deep Learning Training
High-Performance Computing
+ 1 images
Specifications
3rd Gen Intel Xeon Scalable Processors
DDR4 ECC Registered
4U
2x 10GbE BaseT with Intel® X550-AT2 (optional)
2200W (3+1) Redundant High efficiency (Titanium level) Power Supplies
Prices for products are subject to change without notice and DO NOT INCLUDE SHIPPING AND/OR HANDLING CHARGES OR APPLICABLE TAXES. Prices quoted cannot be guaranteed to be accurate until reviewed and accepted by Thinkmate and are intended for same-day purchase. Consequently, prices and special offers are subject to change or expire without notice. Products are available for purchase within the United States. Overseas orders must be prepaid by wire transfer. Thinkmate shall not be responsible for typographical or other errors or omissions regarding prices or any other information. The system configuration information on this page is presented for your convenience only. Due to the vast number of different possible combinations of components and software and the possibility of unexpected unworkable configuration limitations and incompatibilities, in the event that a displayed system configuration must be modified in order to build a properly functioning computer system, Thinkmate shall not be held responsible for any additional costs of reconfiguration of a system design and any additional costs of reconfiguration shall be the responsibility of purchaser. Please note that computer system pictures may not reflect the actual configuration chosen.