NVIDIA Collaborates with Japan on ABCI-Q Hybrid Quantum Supercomputer

NVIDIA is set to collaborate with Japan's National Institute of Advanced Industrial Science and Technology (AIST) to build the nation's cutting-edge hybrid quantum supercomputer, known as ABCI-Q. This project aims to leverage NVIDIA’s advanced HPC and AI GPUs to push the boundaries of quantum computing and high-performance computing.

What Changed Technically?

The key technical advancements in this collaboration include:

• CUDA-Q Integration: NVIDIA will integrate its CUDA-Q platform into the ABCI-Q supercomputer. CUDA-Q is an open-source resource designed to facilitate the development and execution of quantum-classical hybrid applications.
• H100 AI GPUs: The system will be equipped with 2,000 NVIDIA H100 AI GPUs, which are among the most powerful GPUs available today. These GPUs are connected via the latest NVIDIA Quantum-2 InfiniBand interconnects for high-speed data transfer and efficient parallel processing.
• Quantum-Classical Synergy: The hybrid nature of ABCI-Q allows it to perform both classical and quantum computing tasks, making it a versatile tool for research and development in various scientific and industrial domains.

Why It Matters to Practitioners

For researchers and engineers working in the fields of HPC and quantum computing, this collaboration brings several significant benefits:

• High-Performance Simulation: The combination of NVIDIA’s GPUs and CUDA-Q enables researchers to perform complex simulations that are essential for advancing quantum algorithms and applications.
• Ease of Integration: CUDA-Q simplifies the integration of CPUs and GPUs, making it easier for developers to build and deploy hybrid quantum-classical applications.
• Scalability and Efficiency: The use of H100 AI GPUs and Quantum-2 InfiniBand interconnects ensures that the system can handle large-scale computations efficiently, which is crucial for tackling computationally intensive tasks in quantum computing.

Technical Details

• CUDA-Q Platform:
  • Open-Source: CUDA-Q is an open-source framework that supports multiple quantum programming languages and frameworks.
  • Quantum-Classical Applications: It allows developers to write hybrid applications that can run on both classical and quantum hardware, optimizing performance across different computing paradigms.

• NVIDIA H100 AI GPUs:

  • Performance: Each H100 GPU is designed for high-performance AI and HPC workloads, with capabilities in tensor operations, matrix multiplications, and deep learning.
  • Interconnects: The Quantum-2 InfiniBand interconnects provide low-latency and high-bandwidth communication between GPUs, essential for maintaining the performance of large-scale parallel computations.

• System Architecture:

  • Hybrid Design: ABCI-Q is a hybrid system that combines classical HPC resources with quantum computing capabilities.
  • Scalability: The architecture supports scalable deployment, allowing researchers to expand the system as needed to handle more complex tasks.

Impact and Future Prospects

The development of ABCI-Q represents a significant step forward in Japan’s efforts to become a leader in quantum computing and AI. By leveraging NVIDIA's advanced technologies, AIST aims to accelerate research and innovation in areas such as materials science, drug discovery, and optimization problems.

Tim Costa, Director of HPC and Quantum Computing at NVIDIA, highlighted the importance of this collaboration:

"Researchers need high-performance simulation to tackle the most difficult problems in quantum computing. CUDA-Q and the NVIDIA H100 equip pioneers such as those at ABCI to make critical advances and speed the development of quantum-integrated supercomputing."

Conclusion

The partnership between NVIDIA and AIST on the ABCI-Q project is a testament to the growing importance of hybrid quantum-classical systems in advancing scientific research. With its powerful hardware and advanced software, ABCI-Q is poised to become a key resource for researchers in Japan and around the world.