Two processors, in a rack, running on their own
Three things to consider before comparing the specifications below.
Match processor capacity to your applications, concurrent users and virtual machines.
Plan RAM around the services running together, with room for the datasets they use.
Balance fast access with capacity, then plan backups and redundancy for your services.
Side-by-side specification comparison for servers configurations.
| System |  Fusion Rack AI 300 |  Fusion Rack AI 200 |  Fusion Rack Data 300 |  Fusion Rack Data 200 |  Fusion Rack Web 300 |  Fusion Rack Web 200 |  Fusion Rack Fury 300 |  Fusion Rack Fury 200 |
|---|---|---|---|---|---|---|---|---|
| Overview | Four-accelerator AI training node | Two-accelerator AI serving node | Database and storage node | Entry database node | Application and container node | Entry application node | CFD and FEA compute node | Entry CFD compute node |
| Processor | AMD 96 Core Processor | AMD 64 Core Processor | AMD 96 Core Processor | AMD 64 Core Processor | AMD 96 Core Processor | AMD 48 Core Processor | AMD 128 Core Processor | AMD 64 Core Processor |
| Graphics | HPE 192 GB Vram graphics card | HPE 96 GB Vram graphics card | None | None | None | None | PNY 96 GB Vram graphics card | PNY 48 GB Vram graphics card |
| Memory | 768GB High speed memory | 384GB High speed memory | 1152GB High speed memory | 384GB High speed memory | 576GB High speed memory | 384GB High speed memory | 768GB High speed memory | 384GB High speed memory |
| Storage | 65.28TB total | 17.28TB total | 351.36TB total | 40.32TB total | 9.6TB total | 2.88TB total | 4.8TB total | 5.76TB total |
| Peak Power | 3,262W | 2,007W | 1,360W | 1,170W | 1,222W | 858W | 2,036W | 1,401W |
| Deployment | rack | rack | rack | rack | rack | rack | rack | rack |
| Action |
Video memory decides which models fit.
96 Turin cores, 768GB across all 24 memory channels, and four 48GB accelerators pooling 192GB of card memory in one 4U server.
Recommended forAI & Machine Learning
64 cores, 384GB and two 48GB accelerators in the same chassis as the 300, with six bays left free.
Recommended forAI & Machine Learning
Sized by memory speed and your solver licence.
128 cores on 24 filled memory channels, with two 48GB professional cards for GPU-native solvers.
Recommended forComputational Fluid Dynamics
64 cores, 384GB and one 48GB professional card in 4U.
Recommended forComputational Fluid Dynamics
Memory size decides how fast queries run.
1,152GB of error-correcting memory across 24 channels, with 36 drive bays behind it.
Recommended forDatabase & Storage
64 cores, 384GB and ten NVMe drives in 2U: eight for data and two for cache.
Recommended forDatabase & Storage
Web apps and containers, where many small requests run at once.
96 cores and 576GB in 2U.
Recommended forWeb & Application
48 cores at 200W a socket, 384GB, and bonded 10GbE.
Recommended forWeb & Application
A tower that sits beside your desk on a normal power socket and is quiet enough for an office. One person uses it and looks after it.
8 machinesSeveral servers joined by a fast network, for work that no longer fits in one machine. Four designs, each in two sizes, each sized to your site before we quote.
8 machines
