Runtime Optimization Challenges

1
Runtime Optimization Challenges

• Full exploitation of object decomposition
• Load balance within individual phases
• Automatic Identification of phases
• 10K-100K processors
• Inadequate load balancing strategies
• Need topology-aware balancing
• Also fully-distributed decision making
• But based on reasonable global information
• Communication Optimizations
• Identify patterns, develop algorithms suitable
  for each, and implement learning strategies
• Fault tolerance and incremental checkpointing
• Objects as the unit of checkpointing.
• In-memory checkpoints
• Tradeoff between forward-path overhead and
  fault-handling cost

2
Component Frameworks

• Objective
• For commonly used structures, application
  scientists
• Shouldnt have to deal with parallel
  implementation issues
• Should be able to reuse code

• Components and Challenges
• Unstructured meshes Unmesh
• Dynamic refinement support for FEM
• Solver interfaces
• Multigrid support
• Structured meshes Mblock
• Multigrid support
• Study applications
• Particles
• Adaptive mesh refinement
• Shrinking and growing trees
• Applicable to the three above

3
Integration Framework

• Software integration
• Intelligent interface between modules
• Intelligent handling of dynamic behavior
• Adaptivity
• Support for choice of coupling strategies
• Numerical solution transfer
• Across surfaces
• Across volumes
• For multiphysics or mesh refinement
• Support for spawning and coupling subscale
  simulations
• E.g. RocFrac in the middle of RocSolid
• Orchestration of multi- module codes
• Separating flow of control from flow of data
• Challenge
• Flow of control spawns sets of virtual processors

4
(No Transcript)
5
Performance Analysis

• Instrumentation, visualization and analysis
• Current systems and strengths
• Pablo Source level interface, automatic
  instrumentation, and immersive visualization
• Paragraph Multiple intuitive visualizations
• Projections Object based feedback, detailed user
  events, automatic expert analysis
• Challenge
• Bring these ideas to fruition
• Integrate their use in a performance framework
• Develop performance process for software
  development
• So performance issues are not an afterthought