Nvidia pronounces advances in scientific high-performance AI and quantum supercomputing

Nvidia Corp. at this time detailed how its chips are accelerating innovation within the high-performance computing trade for synthetic intelligence-powered methods that can energy scientific advances with supercomputing.

The chipmaker additionally introduced efforts to speed up its quantum computing efforts at nationwide quantum computing facilities all over the world, which can embrace putting in the corporate’s open-source CUDA-Q platform at worldwide websites.

Excessive-performance computing is on the heart of superior scientific analysis and helps unlock vital insights concerning the world at large scales, particularly with AI simulation wanted for atmospheric and oceanic fashions. Excessive computing energy can be required for supplies and life sciences, corresponding to biopharmaceuticals and drug discovery,  the place the computational wants of complicated equations and modeling can rapidly outpace classical methods.

As we speak, Nvidia introduced 9 new supercomputers worldwide which might be utilizing Grace Hopper Superchips for scientific discovery and innovation – in response to the corporate, mixed the methods characterize 200 exaflops, or 200 quintillion calculations per second, of computational AI processing energy.

Among the many new supercomputers coming on-line is the EXA1-HE in France, from CEA, the French Different Energies and Vitality Fee and Eviden, an Atos Group firm, which was introduced in April. It’s primarily based on Eviden’s BullSequana XH3000 know-how. The structure gives a brand new warm-water cooling system and the EXA1-HE is provided with 477 compute nodes primarily based on Grace Hopper.

Different supercomputers coming on-line embrace Helios at Educational Laptop Centre Cyfronet, in Poland; Alps on the Swiss Nationwide Supercomputing Centre from Hewlett-Packard Enterprise; Jupiter on the Jülich Supercomputing Centre in Germany; DeltaAI on the Nationwide Middle for Supercomputing Functions on the College of Illinois Urbana-Champaign; and Miyabi at Japan’s Joint Middle for Superior Excessive-Efficiency Computing.

“AI is accelerating analysis into local weather change, rushing drug discovery and resulting in breakthroughs in dozens of different fields,” stated Ian Buck, vice chairman of hyperscale and HPC at Nvidia. “Nvidia Grace Hopper-powered methods have gotten an important a part of HPC for his or her capability to remodel industries whereas driving higher vitality effectivity.”

Nvidia accelerates quantum computing worldwide

Nvidia is accelerating quantum computing analysis by linking its CUDA-Q platform to supercomputer websites in Germany, Japan and Poland to energy the quantum processing items inside their high-performance computing methods.

QPUs work because the “brains” of quantum computer systems the place all of the quantum processing occurs. They use the conduct of particles corresponding to electrons or photons to make calculations otherwise than conventional processors utilizing qubits, not like a regular bit the place data is usually a 1 or a 0, a qubit can be indeterminate, the place it may doubtlessly be a 1 or a 0. Consequently, quantum processors could make calculations a lot quicker than conventional processors. This additionally signifies that algorithms produced for quantum processors should handle these indeterminate “potential” states, and on the identical time, they are often noisy and error-prone.

Germany’s Jülich Supercomputing Centre at Forschungszentrum Jülich is putting in a QPU constructed by IQM Quantum Computer systems to connect to its Jupiter supercomputer. The ABCI-Q supercomputer in Japan is attaching a QPU from QuEra. Poland’s Poznan Supercomputing and Networking Middle lately put in two QPUs constructed by Orca Computing. 

CUDA-Q is an open-source and QPU-agnostic accelerated supercomputing platform designed to permit for the event of open programming fashions throughout methods between totally different graphical processing items and central processing items. Utilizing CUDA-Q totally different forms of {hardware} and software program might be simply scaled and related and emulated with a quantum useful resource in order that builders can design functions. 

PSNC’s quantum processing items will be used to discover life sciences, chemistry and machine studying with two PT-1 quantum photonics methods. This signifies that it makes use of single photons, or packets of sunshine, at telecommunication frequencies as qubits. This makes it simpler to distribute and hyperlink components collectively utilizing commonplace telecom parts. 

“Our collaboration with Orca and Nvidia has allowed us to create a singular setting and construct a brand new quantum-classical hybrid system at PSNC,” stated Krzysztof Kurowski, chief know-how officer and deputy director of PSNC. “The open, straightforward integration and programming of a number of QPUs and GPUs effectively managed by user-centric providers is essential for builders and customers.”

The QPU built-in with ABCI-Q in Japan will use Rubidium atoms managed by laser mild as qubits, these are the identical sort of atoms utilized by precision atomic clocks. The researchers on the Nationwide Institute of Superior Industrial Science and Know-how will use it to analyze quantum functions in AI, vitality and biology. 

The QPU attached with Jupiter is constructed with superconducting qubits, or digital resonant circuits, that are designed to be manufactured to behave as synthetic atoms at low temperatures. The crew there’ll use it to develop quantum functions for chemical simulations and pave the way in which to display how classical computer systems might be accelerated by quantum computer systems.

Picture: Nvidia