Be part of our day by day and weekly newsletters for the most recent updates and unique content material on industry-leading AI protection. Be taught Extra
Nvidia it’s working with Google Quantum AI to speed up the design of its next-generation quantum computing units utilizing simulations powered by Nvidia.
Google Quantum AI is utilizing the hybrid quantum-classical computing platform and the Nvidia Eos supercomputer to simulate the physics of its quantum processors. This may assist overcome the present limitations of quantum computing {hardware}, which may solely run a sure variety of quantum operations earlier than computations should stop, as a consequence of what researchers name “noise.”
“The event of commercially helpful quantum computer systems is just doable if we are able to scale
up quantum {hardware} whereas preserving noise in verify,” mentioned Guifre Vidal, analysis scientist from
Google Quantum AI, in a press release. “Utilizing Nvidia accelerated computing, we’re exploring the noise
implications of more and more bigger quantum chip designs.”
Understanding noise in quantum {hardware} designs requires advanced dynamical simulations able to absolutely capturing how qubits inside a quantum processor work together with their surroundings.
These simulations have historically been prohibitively computationally costly to pursue. Utilizing the CUDA-Q platform, nonetheless, Google can make use of 1,024 Nvidia H100 Tensor Core GPUs on the Nvidia Eos supercomputer to carry out one of many world’s largest and quickest dynamical simulation of quantum units — at a fraction of the price.
“AI supercomputing energy can be useful to quantum computing’s success,” mentioned Tim Costa, director of quantum and HPC at Nvidia, in a press release. “Google’s use of the CUDA-Q platform demonstrates the central position GPU-accelerated simulations have in advancing quantum computing to assist remedy real-world issues.”
With CUDA-Q and H100 GPUs, Google can carry out absolutely complete, practical simulations of units containing 40 qubits – the biggest carried out simulations of this sort. The simulation methods supplied by CUDA-Q imply noisy simulations that might’ve taken per week can now run in minutes.
The software program powering these accelerated dynamic simulations can be publicly out there within the CUDA-Q platform, permitting quantum {hardware} engineers to quickly scale their system designs.
