Back to Research

University of Alberta  

Project 1152 - Effective Support for Multi-Processor Execution of Enterprise Applications

SHARE THIS POST

Running from 2022 to present

Effective Support for Multi-Processor Execution of Enterprise Applications

Large software projects often have significant compilation times due to the complexity of their code, and their need for performant code that requires numerous optimizations. Targeting compilation to specific hardware configurations can greatly benefit the capabilities of performance optimizations, but also reduces the portability of the compiled program. As a result vendors often compile in a generic manner that works on multiple configurations, in the process restricting the room for optimizations. To better address this problem, we propose a method wherein we identify functions in a program that have the possibility of benefiting from target dependent optimizations. With that knowledge we then implement a JIT compiler that will recompile those function's code when the program is run, using runtime information to produce highly optimized code that performs better than what was possible ahead of time.

Learn More about the Research Team.  

Explore the product that harvests this research results  

Research team:

  • PI: Prof. J. Nelson Amaral, University of Alberta
  • Student: Tyler Gobran, University of Alberta
  • Student: João Carvalho, University of Alberta
  • Student: Dhanrajbir Singh Hira, University of Alberta
  • Student: Ivan Korostelev, University of Alberta
  • IBM Project Lead (RCL): Kit Barton, IBM
  • IBM Manager (RCM): Wang Chen, IBM
  • IBM Sponsor (RCS): Kevin Stoodley, IBM
  • IBM Contributor (RCC): Nemanja Ivanovic, IBM

Institution:

University of Alberta   

SHARE THIS POST