design features
Maximum acceleration in one Hyperion node
AxiForge Max is the largest single-node training system in the current Hyperion range. Four professional 96GB accelerators share one enterprise server with 128 AMD EPYC cores and 1.15TB of ECC memory.
- One Supermicro AS-4125GS-TNRT / CSE-418G2TS 4U rackmount node
- 128 CPU cores and 1.15TB ECC DDR5 memory
- Four professional NVIDIA GPUs with 384GB aggregate VRAM
- Designed for AI training, engineering AI and applied research
Hyperion X1460 — AxiForge Max AI Training System description
Cluster-class capability without a fabric to manage
Hyperion X1460 concentrates the accelerator, CPU and memory resources needed for large-model fine-tuning, generative engineering, simulation-surrogate training and confidential research pipelines into a single managed node. It can also become the first node of a larger Hyperion platform when requirements expand.
The online configurator provides live indicative configuration pricing and validates compute, memory, storage, networking, rack, power and compliance selections together. Final price is subject to component availability, export compliance and formal quotation.
Technical library
Explore the engineering behind this Hyperion system
Read workload guides, software-stack guidance, application notes and platform briefs written for deployment and validation.
Contact us at sales@axiotech.co.uk and we’ll be happy to advise.