Fusion Rack AI 300
- Processor
- AMD 96 Core Processor
- Memory
- 768GB High speed memory
- Video memory
- HPE 192 GB Vram graphics card
Why this machine
96 Turin cores, 768GB across all 24 memory channels, and four 48GB accelerators pooling 192GB of card memory in one 4U server. Eight NVMe bays hold your datasets, because at this scale data loading falls behind before the cards do.
Made for your workflow
Performance at a glance
Explore the hardware behind your next project. Real capacity, explained at a glance.
- Processing power
- 96 cores
- 2 processors · 192 threads
- CPU cores handle calculations, simulations and tasks running in parallel.
- Working memory
- 768GB
- DDR5 · ECC error correction
- RAM holds the data your applications are using, from project files to active datasets.
- Graphics & acceleration
- 4 GPUs
- 4 × L40S 48GB (HPE OEM) · 48GB per card
- Dedicated GPU memory supports graphics, rendering and compatible AI workloads.
- Local storage
- 65.28TB
- 10 drives · Includes NVMe SSD · Total drive capacity
- Storage gives your projects, datasets and results a home between sessions.
Hardware platform
From your data to your next result
- StoreFiles & datasets
- LoadActive working data
- ProcessTasks & calculations
- AccelerateGPU workloads
What it is for
We recommend this machine for one workload. Each one explains what the work actually stresses.
Included with every system
- 72 hours of burn-in before it leaves us
- Delivered across Pakistan from Islamabad, by courier
- Warranty handled here rather than shipped abroad
- One invoice and one point of contact for the whole machine
Measured
- Threads
- 192
- Accelerators
- 4
- BF16 throughput
- 1448 TFLOPS
- Storage
- 65.28TB
- Peak draw
- 3,262W
- At 230V
- 14.2A
- Heat rejected
- 9,458 BTU/hr
- Deployment
- 4U rack
Every figure above is computed from the parts list, using the same summary the configurator runs. Power is the sum of the parts at their rated draw with the usual headroom, not a nameplate figure.
