Adaptive Parallelization Strategies using Data-driven Objects

1
Adaptive Parallelization Strategies using
Data-driven Objects

• Laxmikant Kale
• David Padua

2
Outline

• Quench and solidification codes
• Coarse grain parallelization of the quench code
• Adaptive parallelization techniques
• Dynamic variations
• Adaptive load balancing
• Finite element framework with adaptivity
• Preliminary results

3
OpenMP
4
(No Transcript)
5
(No Transcript)
6
Coarse grain parallelization

• Structure of current sequential quench code
• 2-D array of elements (each independently
  refined)
• Within row dependence
• Independent rows, but
• share global variables
• Parallelization using Charm
• 3 hours effort (after a false start)
• about 20 lines of change to F90 code
• A 100 line Charm wrapper

7
Performance results
Contributors Engineering N. Sobh, R.
Haber Computer Science M. Bhandarkar, R.
Liu, L. Kale
8
Adaptive Strategies

• Advanced codes model dynamic and irregular
  behavior
• Solidification adaptive grid refinement
• Quench
• Complex dependencies,
• Parallelization within elements
• To parallelize these effectively,
• adaptive runtime strategies are necessary

9
Multi-partition decomposition using objects

• Idea decompose the problem into a number of
  partitions,
• independent of the number of processors
• Partitions gt Processors
• The system maps partitions to processors
• The system should be able to map and re-map
  objects as needed

10
Charm

• A parallel C library
• Supports data driven objects
• singleton objects, object arrays, groups,
• Many objects per processor, with method execution
  scheduled with availability of data
• System supports automatic instrumentation and
  object migration
• Works with other paradigms MPI, openMP, ..

11
Data driven executionin Charm
Scheduler
Scheduler
Message Q
Message Q
12
Load Balancing Framework

• Aimed at handling ...
• Continuous (slow) load variation
• Abrupt load variation (refinement)
• Workstation clusters in multi-user mode
• Measurement based
• Exploits temporal persistence of computation and
  communication structures
• Very accurate (compared with estimation)
• instrumentation possible via Charm/Converse

13
Object balancing framework
14
Utility of the framework workstation clusters

• Cluster of 8 machines,
• One machine gets another job
• Parallel job slows down on all machines
• Using the framework
• Detection mechanism
• Migrate objects away from overloaded processor
• Restored almost original throughput!

15
Higher level framework
Automatic Conversion from MPI
Cross module interpolation
Structured
FEM
MPI-on-Charm
Irecv
Frameworkpath
Load database balancer
Migration path
Charm
Converse
16
Example application

• Crack propagation
• (P. Geubelle et al)
• Similar in structure to Quench components
• 1900 lines of F90
• Rewritten using FEM framework in C
• 1000 lines of C code
• Parallelization completely by the framework

17
Crack Propagation code, C version, with 70k
elements
18
Crack propagation preliminary results
Artificially emulates load imbalance created by
extra work near the crack (e.g. due to adaptive
refinement) Data obtained on 8 processors of
Origin 2000
19
Summary and Planned Research

• Use the adaptive FEM framework
• To parallelize Quench code further
• Quad tree based solidification code
• First phase parallelize each phase separately
• Parallelize across refinement phases
• Refine the FEM framework
• Use feedback from applications
• Support for implicit solvers and multigrid

20
(No Transcript)