By linking quantum and GPU computing, researchers can now run large-scale simulations and hybrid workloads extra effectively.
Quantum computer systems compute with qubits, completely in another way from regular computer systems, however these qubits are unstable and make frequent errors. To repair them, quantum techniques need assistance from highly effective classical computer systems that may shortly run correction algorithms. The issue is that present connections between quantum processors and classical techniques are too gradual for this real-time correction.
NVIDIA’s new NVQLink goals to unravel this drawback. It’s a high-speed knowledge hyperlink that instantly connects quantum processors to GPU-based supercomputers. The hyperlink permits knowledge to maneuver between the 2 techniques virtually immediately, making it attainable to manage and proper qubits as they work.
Right here’s the way it capabilities: throughout a calculation, the quantum processor sends knowledge about its qubits to the GPU system. The GPUs then analyse this info, calculate any corrections wanted, and ship these corrections again by means of NVQLink. This occurs repeatedly, serving to the quantum system keep secure and correct.
NVQLink works with NVIDIA’s CUDA-Q software program, which lets builders run quantum and classical computing collectively throughout the identical program. This integration helps create hybrid techniques that may deal with advanced issues in supplies science, chemistry, and superior simulations a lot sooner than conventional computer systems alone.
The setup combines the strengths of each techniques. The quantum processor handles quantum-specific duties like molecular simulations or encryption, whereas the GPUs deal with quick knowledge processing and management duties. This “hybrid” method helps researchers use quantum computer systems extra successfully and scale them for bigger experiments.
NVIDIA developed NVQLink with a number of U.S. analysis centres, together with Brookhaven, Fermilab, Lawrence Berkeley, Los Alamos, Oak Ridge, Pacific Northwest, Sandia, and MIT Lincoln Laboratories. The expertise additionally helps many quantum {hardware} and controller builders, making it simpler to attach completely different quantum techniques.
