SupermicroGPU server
8U 8-GPU HGX server
AI training · HPC · Inference
- Form factor
- 8U rackmount
- GPUs
- 8× SXM5 on HGX baseboard
- Sockets
- 2
Dual-socket 4U platform for up to 8 double-width PCIe accelerators.
A 4U dual-socket platform that accepts up to eight double-width PCIe accelerators such as the L40S. It suits inference, rendering and virtualisation deployments where standard PCIe cards are preferred to an HGX baseboard.
Inference4U PCIe GPU server is sourced for inference deployments within servers, configured to the host platform and quantity specified in your requirement.
Virtualisation4U PCIe GPU server is sourced for virtualisation deployments within servers, configured to the host platform and quantity specified in your requirement.
Workstation4U PCIe GPU server is sourced for workstation deployments within servers, configured to the host platform and quantity specified in your requirement.
Accelerator model, CPU, memory and storage are specified per requirement. Confirm per-slot power budget for the intended GPU.
AI training · HPC · Inference
Virtualisation · Storage · Inference
Tell us your requirement and we'll reply with sourcing options and lead time.