UC Davis Research Team Enables Extreme-Scale Visualization using oneAPI
October 26, 2021
The VIDI research group at UC Davis has established a oneAPI academic Center of Excellence focusing on the innovation of extreme-scale data visualization. With oneAPI, the center will build, test, and deploy performance-driven, cross-architecture applications that scale across CPUs and GPUs. The center will use Intel oneAPI Rendering Toolkit to develop high-performance, high-fidelity data visualization and analysis applications and solutions on Intel CPU and accelerator architectures, including upcoming Intel Xe GPUs.
The VIDI research group led by Professor Kwan-Liu Ma is among the leaders in scientific visualization research. Their work is largely driven by both simulation and experimental sciences in a highly interdisciplinary setting. In particular, they have been working closely with scientists who are developing large-scale simulations to study complex physical phenonmena and chemical processes for new discoveries and advancements. These state-of-the-art simulations can generate hundreds of petabytes of data that are infeasible to write out to disk at sufficiently frequent intervals for reliable post-hoc analysis. For the scientists to fully exploit the simulations at the extreme scale, in situ visualization support is in need to identify, track, and understand features of interest in the modeled phenomenon without sacrificing simulation fidelity, nor inducing excessive computing and storage costs. VIDI is known for its leadership and innovations in the development of in situ data visualization and analysis solutions, especially based on several novel approaches to spatiotemporal feature tracking and rendering of data at extreme-scale in a massively parallel computing environment. Through oneAPI, VIDI will develop in situ visualization software technologies that are highly portable to take advantage of future heterogeneous computer architectures, which will greatly enhance scientists' ability to explore and discover.
"The UC Davis team is committed to developing high-performance visualization software solutions through the use of oneAPI technologies and educating best practices to both the visualization research and scientific user communities," says Kwan-Liu Ma, Distinguished Professor of Computer Science.
"Intel is pleased to support Professor Kwan-Liu Ma's groundbreaking, world-recognized research in visual computing. His team's leadership in simulation and experimental sciences, especially in feature-tracking as we get to exascale, becomes more important than ever in maximizing performance of mixed architecture systems and platforms. We look forward to collaborating with Professor Ma to utilize the open, flexible approach of oneAPI cross-architecture programming to accelerate adoption of new innovative architectures with uncompromised performance - overcoming the constraints of proprietary programming models. Together, we continue the journey leaving no transistor and no pixel left behind," says Jim Jeffers, senior principal engineer and senior director of Intel Advanced Rendering and Visualization.