Zike Xu

Graduate in Computer Science and Technology

About

Hi! I am a first-year graduate studying Computer Science at ShanghaiTech University. I was an exchange at University of California, Berkeley.

I specialize in Computer Graphics and Aerodynamics, with particular interests in the research fields of Simulation and Rendering , e.g. physically-based fluid simulaiton and light transport simulation. I am currently conducting research at FLARE lab, under the guidance of Prof. Xiaopei Liu.

Projects

AMBER (Aerodynamics Multi-GPU Block-based solvER)

High-Performance LES Simulator for Aerodynamics Flow

  • Developed a CUDA-accelerated Large Eddy Simulation (LES) solver for unsteady turbulent flows, implementing an optimized Lattice Boltzmann Method (LBM) kernel achieving near-peak memory bandwidth utilization.
  • Engineered multi-GPU, multi-resolution and immersed boundary handling capability that supports up to $8$ GPUs on a node, enabling simulations of complex geometries at Reynolds numbers up to Re~1e6 under low-speed flow regimes with fidelity.

FLux (FLuid Under optiX)

High-Performance Physically-Based Renderer with OptiX

  • Implemented a JITed GPU render engine with a focus on extreme efficiency. Making extensive use of OptiX 8’s capability, it achieves the strict criterion of zero-overhead principle through JITed kernels on GPU: enabled new features don’t introduce performance regression while operating at maximum efficiency.
  • Optimized the clean C++ 20 codebase with efficiency in mind, common features like SER-accelerated megakernel path tracing, quasi-Monte Carlo sampling and comprehensive set of materials including the multi-layer ones are not sacrificed.

Education

ShanghaiTech University

Bachelor of Engineering in Computer Science and Technology

Aug. 2020 - Jun. 2024

University of California, Berkeley

Exchange student in Computer Science and Technology

Jan. 2023 - May 2023

ShanghaiTech University

Master of Science Candidate in Computer Science and Technology

Sep. 2024 - Jun 2027 (expected)