This project aims at increasing floating point performance by the following improvements:

  • In Simplifier and Value Propagation, less work is done on Floating Point nodes as compared to integral nodes, which is to be resolved.
  • Implement support for Fused Multiply-Add (FMA) instruction set in OMR x86 backend.
  • Add IL opcodes for FMA variants.
  • x87 is to be removed and replaced by SSE where required.

Organization

Student

Shivam Mittal

Mentors

  • Daryl Maier
close

2018