Nvidia, the darling of excessive efficiency computing (HPC), is bringing new consideration to quantum computing.
The corporate has launched its Nvidia Quantum Optimized System Structure, or QODA. This hybrid platform is designed to make quantum computing extra accessible by enabling programming of each quantum functions and classical functions in a single, consolidated atmosphere. In line with Nvidia, it is geared toward dashing breakthroughs in quantum analysis and growth throughout AI, HPC, well being, finance and different disciplines.
The intention is to make QODA be for quantum computing simply as CUDA is for GPU computing – an trade customary. (CUDA is a C-like code for writing specialised HPC and AI functions that run on Nvidia GPUs.)
Nvidia mentioned HPC and AI builders can use QODA so as to add quantum computing to present functions, leveraging each quantum processors in addition to simulated future quantum machines utilizing Nvidia DGX programs and Nvidia GPUs obtainable in scientific supercomputing facilities and public clouds.
This isn’t Nvidia’s first dance within the quantum computing area. A few yr in the past, the corporate launched cuQuantum, a software program growth equipment (SDK) for accelerating quantum workflows utilizing the Tensor Cores in its GPUs together with numerous libraries and instruments optimized for such jobs as quantum circuit simulations.
In asserting the brand new structure, Nvidia has introduced collaborations with a number of quantum computing companies, most with a Q of their title: {hardware} distributors IQM Quantum Computer systems, Pasqal, Quantinuum, Quantum Brilliance and Xanadu; software program suppliers QC Ware and Zapata Computing; and supercomputing facilities Forschungszentrum Jülich, Lawrence Berkeley Nationwide Laboratory and Oak Ridge Nationwide Laboratory, which is fascinating as a result of ORNL is an all-AMD store.
What’s quantum computing?
The processes have superior, however the fundamental construction of computing has not modified because it was invented. Knowledge is represented at its most elementary state as bits, 0 or 1. Quantum computing makes use of one thing known as a quibit, which might symbolize a 0, a 1, or any proportion of 0 and 1 in superposition of each states. A quibit might be 1/4 0 and three/4 1, for instance.
This implies a number of issues. First, a lot better velocity. Quantum computing can course of information at as much as 1,000 occasions sooner than customary binary computer systems. You’re not going to make use of a quantum pc to run Microsoft Excel. You’re going to make use of a quantum pc to do climate simulations and drug testing.
Second, quantum computing has no compatibility with present functions. You’re not simply going to rewrite or recompile an utility on a quantum pc. You must write the entire thing over from scratch to utterly make the most of the brand new structure. Nobody goes to search out that interesting, and that’s what Nvidia is attempting to handle.
Happily, it’s not an both/or state of affairs. Functions that will probably be accelerated by quantum processing items (QPU) will probably be hybrid workloads that leverage a normal supercomputing structure for big components of an utility, whereas probably the most important components are accelerated by a quantum system.
Copyright © 2022 IDG Communications, Inc.
