Mission

The Global Computing Lab (GCLab), headed by Dr. Michela Taufer, targets large-scale, heterogeneous computer systems and their application to the sciences.

Our research focuses on exploring, designing, and implementing alternative, more efficient computational paradigms for large-scale scientific simulations on hybrid computing systems including grid and volunteer computing, multi-core architectures, and graphics processing units (GPUs).

Our main challenges are:

  • The ability to model biological systems with computational algorithms that dynamically adapt to simulation results and performance collected at run-time on hybrid systems.
  • The ability to assure that these simulation algorithms can be executed in the "required" time accurately.

The first point refers to the adaptive calibration of prediction models based on the complexity and characteristics of the biological systems as well as, on the availability and reliability of hybrid computational resources. The second point refers to having the necessary computational resources (computing cycles, memory, network, etc.) to accomplish the simulation work in the required time and being able to trust the simulation results: these results can be affected by e.g., hardware malfunctions, incorrect software modifications, and floating-point arithmetic with insufficient precision.