NVIDIA, IBM and Fermilab plan quantum-classical platform
NVIDIA, IBM and Fermilab plan to build a modular research platform linking quantum processors with accelerated classical computing, while NVIDIA is also introducing a faster quantum error-correction decoder and expanding quantum access through supercomputing software.
The platform is intended to help researchers study error-correction codes, decoding algorithms and applications combining quantum and classical systems. NVIDIA DGX B200 systems are to be installed alongside IBM quantum computers, including systems using IBM Quantum Nighthawk, at Fermilab’s Superconducting Quantum Materials and Systems Center. The organizations said the work will also examine how classical computing can scale alongside interconnected quantum processors.
NVIDIA says its TCBPJOBS algorithm, available through the CUDAQ QEC 0.9.0 library, can accelerate quantum error-correction research workloads by up to 13,000 times compared with leading existing algorithms. The company says one computation used to study new quantum-computing designs could be reduced from 12 weeks to 45 minutes. Those figures are company-reported performance claims, not evidence that quantum processors have achieved fault-tolerant...