Ocean Benchmark (MPI) - Finite Difference

Code provided by Northrop-Grumman (NAVO)

Benchmark Ocean Model

Optimized Performance

Benchmark Ocean Model

Optimized Performance

• Out of the box
  ~ 70MF / 4 nodes

• Cache block two main loops, split come communication
  ~ 175MF

• Inter-procedure analysis and eliminate 2 matrix copies
  ~ 195 MF

• Eliminate communication associated with extra copies
  ~ 235 MF

• Loop jam
  ~ 220 MF

• Cache block jammed loop
  ~ 355 MF

• Fix Legion-MPI bug
  ~ 480 MF

[Centurion Overview] [Applications] [Photos]

[Index Page] [Overview] [Project Status] [Download Legion] [Security] [Prototypes] [Documents] [Documentation] [Presentations] [Promotional Material] [Workshops] [Contact Information] [Team Members] [Job Opportunities] [Access Statistics] [Centurion]

legion@virginia.edu
http://legion.virginia.edu/