Application Template:
1) Year 1: 
Establish Procedure
Do 2 Simple Cases – Syracuse Black Hole, Maryland ?
Establish collaboration with Parkbench activity
2) Year 2: 
2 More “simple cases” including adaptive but still “pure” data parallel
Do first complex distributed Application (John Hopkins?)
3) Year 3:
Evaluate templates on tools and develop at least one more Distributed Application

Problem Solving Environment – PETASIM Activity
1) Year 1:
Develop simple standalone PETASIM with user control and evaluate on 
application templates, “petaflop/simple kernel parkbench kernels”
Develop Java Interface to PETASIM which can be part of the TANGOsim PSE 
frontend
Develop common API of Runtime/Application code to PETASIM and more 
sophisticated simulaters
2) Year 2: 
Extend PETASIM to adaptive applications
Develop and test User directives to drive interpretation of Performance simulators 
(both PETASIM and sophisticated case) by runtime
Ongoing use and evaluation
3) Year 3:
Extend PETASIM to distributed applications
Ongoing Use and Evaluation

Problem Solving Environment – PCRC Runtime Activity
1) Year 1:
Design Link of PCRC to Performance API described above
Demonstrate key functions of PCRC linked to PETASIM 
2) Year 2:
Enhance PCRC functions with performance sensitive algorithms for selected 
target machines and evaluate
3) Year 3:
Full implementation of PCRC with performance sensitive algorithms

Application Testbed
1) Year 1:
Link TANGOsim to performance system with one module PETASIM and others 
executing applications and users in computational steering environment – thi 
would not be in context of a real application
2) Year 2:
Extend linkage to real applications – especially the John Hopkins Testbed
Support migration of applications and other identified features
3) Year 3:
Evaluate and Extend testbed so that it is example of a PSE (Note TANGOsim is 
essentially PSE infrastructure) with performance sensitive runtime with mix of 
user and automatic control