Beijing Normal University CPU+GPU High-Performance Server

[Customer Case] Quantum State Simulation High-Performance Computing Platform

I. Customer Background

A leading physics research team specializing in quantum state simulation and algorithm development.

II. Core Requirements

1. CPU: Intel platform, multi-core architecture; maximize core count within budget

2. Memory: ≥ 2 TB

3. GPU: Total CUDA cores ≥ 10000

4. Form Factor: Rack Server

5. Budget: 25 ten thousand yuan

III. Challenges Ahead

• Compute bottleneck: Traditional CPUs are too slow to meet the demands of large-scale quantum state simulation.

• Stability: Must run complex tasks continuously for extended periods, requiring extremely high hardware reliability.

IV. Recommended Solution: TRY SR212(P83-SSX1)

Hardware Configuration:

- CPU: 80 cores / 160 threads, all-core boost frequency up to 2.98 GHz

- Memory: 2 TB DDR4 3200 ECC REG

- GPU: NVIDIA RTX 6000 Ada 48GB, CUDA Cores: 18176

Key Features:

✓ Dual Intel platform: maximum core count within budget

✓ 2TB large memory for quantum state vector storage

✓ Single card with 1.8 million CUDA cores—far exceeding requirements and supporting mixed-precision computing.

5. Implementation Results

• Quantum state simulation speed increased by 300%

• Reduced experiment cycle from 2 weeks to 3 days

• Successfully published 3 papers in Nature sub-journals

• Platform has operated without failure for more than 2000 hours

• Maximum simulated qubits increased from 30 to 36 Qubit

6. Customer Feedback

"The new platform accelerates simulation speeds by 3x, reducing a two-week workload to just three days. This allows us to validate more hypotheses in less time, ultimately leading to publication in a Nature sub-journal." — Project Lead