1. Basic Configuration

• Model: SuperCompute SC-500
• Chassis: 2U Rackmount (482×900×89mm)
• Weight: 24kg (base), ≤32kg (max)

2. Core Hardware

• CPU:

• Model: AMD EPYC 7763
• Cores/Threads: 64C/128T
• Freq: 2.45GHz base / 3.5GHz max
• Cache: 256MB L3
• TDP: 280W
• PCIe Lanes: 128×PCIe 4.0

• Memory:

• Capacity: 256GB (8×32GB)
• Type: DDR4-3200 8-channel RDIMM
• Max Expansion: 16 slots, up to 4TB
• Bandwidth: 204.8GB/s

• Accelerators:

• Model: 2×NVIDIA A100 40GB
• CUDA Cores: 6,912 (per GPU)
• VRAM: 40GB HBM2 (1,555GB/s)
• Interconnect: NVLink (600GB/s)

• Storage:

• Seq. R/W: 7GB/s / 4.2GB/s
• Random R/W: 1M/360K IOPS
• OS Drive: 2×1.92TB Intel P5510 NVMe SSD
• Optional: 12×8TB HDD (Ceph)

3. Networking & Expansion

• High-Speed Network:
• Mellanox ConnectX-6 100Gb EDR InfiniBand (<0.7μs latency)
• RDMA & GPUDirect support
• Expansion Slots:
• 2×PCIe 4.0 x16 (for GPUs)
• 1×PCIe 4.0 x8 (for NVMe)

4. Cooling & PSU

• Cooling:
• Redundant fans (N+1)
• Optional liquid cooling (direct-to-chip)
• Power Supply:
• 1600W 2+2 redundant (80Plus Titanium)
• Input: 200-240V AC (50/60Hz)

5. Performance

• Benchmarks:
• LINPACK Efficiency: 94.7% (FP64)
• HPL (Rmax): 5.2 TFLOPS
• AI Training (ResNet-50): 1,500 images/sec
• Workloads:
  ✓ Computational Fluid Dynamics (CFD)
  ✓ Molecular Dynamics (NAMD/GROMACS)

6. Use Cases

• Recommended:
  ✓ Scientific Computing (Climate/Astrophysics)
  ✓ AI Training (LLM/Computer Vision)
  ✓ Quantitative Finance (HFT Backtesting)
• Deployments:
  ▶ National Supercomputing Center (200+ nodes)
  ▶ Top Pharma Molecular Simulation

7. Service

• Warranty: 3-year OEM (incl. liquid cooling)
• Options:
  ✓ HPC Cluster Deployment
  ✓ Performance Tuning (MPI/OpenMP)