TEL AVIV, Israel, June 10, 2025 (GLOBE NEWSWIRE) -- LightSolver, inventors of a new laser-based computing paradigm, and HLRS (High-Performance Computing Center Stuttgart) today announced the joint presentation of their co-authored, peer-reviewed research paper, “Exploring QUBO on LPUs for Engineering,” at this week’s ISC High Performance 2025 conference in Hamburg, Germany. The presentation will take place this Friday, June 13th during the 4th HPC on Heterogeneous Hardware (H3) workshop, on the first floor of Hall X6 from 2:00-6:00pm.
With HPC simulations demanding immense computational power, emerging specialized processors like LightSolver’s Laser Processing Unit™ (LPU) offer a promising path toward reducing total time to solution and improving energy efficiency. This is reflected in the research paper, which explores the use of the LPU to address mesh decomposition, a common task in Computer-Aided Engineering (CAE). It will be presented by Dr. Dan Gluck, senior algorithm researcher for LightSolver, and Dr. Johannes Gebert, deputy director at HLRS and group lead for future computing.
Mesh decomposition is the process of breaking down a complex 3D mesh (a collection of vertices, edges, and faces that define the shape of a 3D object) into simpler subcomponents. This technique is widely used in computer graphics, geometry processing, 3D modeling, finite element analysis, and robotics. For their research, LightSolver and HLRS used Direct Tensor Computation, a biomechanical application in which mathematical operations are performed directly on multi-dimensional arrays called tensors, rather than being broken down into simpler components. The team converted the mesh decomposition into a QUBO (Quadratic Unconstrained Binary Optimization) formulation and ran it on the LPU emulator, developing a methodology and proving that mesh decomposition could be addressed by an optical computing platform.
“We are proud to present our joint research with HLRS, which highlights how optical computing can address common bottlenecks in engineering software,” said Ruti Ben Shlomi, CEO of LightSolver. “To crack current performance limits and achieve greater energy efficiency, it is crucial for industry and research to partner and examine novel computing paradigms. This initial paper has demonstrated the feasibility of running a mesh decomposition on our optical processor, the LPU. We’re looking forward to deepening our collaboration with HLRS and presenting practical pathways for integrating specialized processors into heterogeneous computing frameworks.”
The H3 Workshop provides a platform for pioneering work in algorithmic research, software libraries, programming models, and workflows designed for increasingly heterogeneous hardware environments. The presentation by LightSolver and HLRS aligns with the workshop's mission to explore new paradigms and strengthen the bridge between software and specialized hardware in HPC.
About LightSolver
LightSolver is a photonics computing company that is developing an all-optical supercomputer capable of solving complex and large computational problems at the speed of light." Utilizing the interference patterns of lasers, the Laser Processing Unit™ (LPU) can tackle challenges that were previously constrained by the limits of electronics, while fitting into a rack unit and operating at room temperature. Dr. Ruti Ben-Shlomi and Dr. Chene Tradonsky, physicists from the world-renowned Weizmann Institute, founded the company in 2020. More than 2/3 of the team are physics, math and computer science PhDs. LightSolver has secured investment from TAL Ventures, Entree Capital, IBI Tech Fund, Angular Ventures, Maverick, and Artofin. The company has also received a €12.5M grant from the European Innovation Council (EIC) to advance its all-optical supercomputer. Connect with LightSolver @LightSolverCo on X and on LinkedIn. For more information, visit lightsolver.com or email info@lightsolver.com.
Media Contact:
Seth Menacker
Fusion PR
lightsolver@fusionpr.com
