4UV GPU Accelerator
Eletter Product
The 4UV GPU Accelerator connects to any Gen 4 server, including OSS Gen 4 Expansion Optimized Servers (EOS), via two PCIe x16 Gen 4 links. Featuring 10 PCIe 4.0 slots and 4,000 watts of load-sharing power, the 4UV supports up to 8 NVIDIA V100S Tensor Core GPUs and delivers 1,040 teraFLOPS of tensor performance and 65.6 teraFLOPS of double precision performance. The 4UV can also accommodate up to 16 NVIDIA T4 GPUs and delivers 1,040 teraFLOPS of mixed-precision performance. This system is the newest AI on the Fly® product and delivers data center capabilities to HPC and AI edge deployments in the field or for mobile applications. The 4UV is ready-made for AI training and large scale inferencing.
Features:
- 4U rackmount design
- 10 PCIe 4.0 x16 expansion slots
- Two PCIe 4.0 x16 host connections
- Two PCIe copper or fiber cable sets
- Fixed or variable speed fan options
- Two 2000W load-sharing power supplies
- AUX power connectors for GPUs high power AICs
