Static Binary Rewriting:

Rewriting the binary with parallelized loops involves the folllowing procedure:

  1. Shared variables are moved to the shared data area; variable references are changed to refer to the new location.
  2. Private variables are moved to the per-thread data area; variable references are changed to refere to the new location.
  3. Initialization and/or finalization of private variables is arranged.
  4. Parallel loop bodies are abstracted into functions whose addresses are pased to the system scheduler.
  5. Relevant library calls for the underlying parallel platform (e.g., OpenMP) are inserted as needed.

Static Binary Rewriting:

Rewriting the binary with parallelized loops involves the folllowing procedure:

  1. Shared variables are moved to the shared data area; variable references are changed to refer to the new location.
  2. Private variables are moved to the per-thread data area; variable references are changed to refere to the new location.
  3. Initialization and/or finalization of private variables is arranged.
  4. Parallel loop bodies are abstracted into functions whose addresses are pased to the system scheduler.
  5. Relevant library calls for the underlying parallel platform (e.g., OpenMP) are inserted as needed.
© Copyright 2009, All Rights Reserved | Contact Us