1. A High-Level Framework for Parallelizing Legacy Applications for Multiple Platforms Ritu Arora Texas Advanced Computing Center Email: rauta@tacc.utexas.edu

2. Outline Motivation and Goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 2

3. Plenty of Parallel Programming Languages and Paradigms MPI OpenMP CUDA OpenCL Implicitly Parallel Languages (X10, Fortress, SISAL) PGAS languages (UPC, Co-Array Fortran) Offload programming for MIC SHMEM Cilk/Intel Cilk Plus Charm++ HPF Hybrid programming (MPI + OpenMP) 3

4. 4 There is a need to develop a tool (high-level framework) that offers a low-risk way for domain-experts to try HPC but first…

5. … Understanding the Mindset of the User Community is Important “…the history of HPC is littered with new technologies that promised increased scientific productivity but are no longer available.” “A new technology that can coexist with older ones has a greater chance of success than one requiring complete buy-in at the beginning.” “For many frameworks, a significant barrier to their use is that you can’t integrate them incrementally.” “Frameworks provide programmers a higher level of abstraction, but at the cost of adopting the framework’s perspective on how to structure the code.” 5 Source: Understanding the High Performance Computing Community: A Software Engineer’s Perspective, Basili et al.

6. Standard and Non-Standard Steps for Parallelization that are Repeatable Examples of standard steps in developing an MPI application (common in all MPI programs) – Every MPI program has #include "mpi.h" – Every MPI program has MPI_Init and MPI_Finalize function calls Non-standard steps in developing an MPI application – for-loop parallelization, data distribution, mapping of tasks to processes, and orchestration of exchange of messages Steps for splitting the work in a for-loop amongst all the processes in MPI_COMM_WORLD are standard for a given load-balancing scheme 6

7. Goals Develop a high-level framework for semi-automatic parallelization that can leverage the investment made in legacy applications The framework should be built on top of successful programming paradigms like MPI, OpenMP, and CUDA Provide support for incremental parallelism through the framework Abstract the standard and non-standard steps in parallelization 7

8. Outline Motivation and Goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 8

9. How Does the Framework Work? 9

10. Providing Specifications Through Hi-PaL (1) Parallel section begins ( ) mapping is { && in function ( ) } 10 OMP_Parallel { && schedule is && in function ( ) } General Structure of Hi-PaL Code to Generate MPI Code General Structure of Hi-PaL Code to Generate OpenMP Code

11. Providing Specifications Through Hi-PaL (2) A set of Hi-PaL API has been developed for precisely capturing the end-users’ specifications at a high-level 11 Hi-PaL APIDescription ParExchange2DArrayInt(,, ) Exchange neighboring values in stencil-based computations Parallelize_For_Loop where ( ; ; ) Parallelize for-loop with matching condition, stride and initialization statement ReduceSumInt( ) MPI_Reduce with MPI_SUM operation or OpenMP reduction clause with ‘+’ operator; reduced variable is of type integer

12. Parallelizing Poisson solver (1) 12 1. //other code 2. NTIMES = atoi(argv[3]); 3. a = allocMatrix (a, M, N); 4. b = allocMatrix (b, M, N); 5. f = allocMatrix (f, M, N); 6. start = 0; 7. //other code 8. printMatrix (a, M, N); 9. t1 = gettime(); 10. for (k = start; k = tolerance; k++) { 11. b = compute(a, f, b, M, N); 12. ptr = a; 13. a = b; 14. b = ptr; 15. norm = normdiff(b, a, M, N); 16. } 17. t2 = gettime();//other code Code snippet of serial Poisson Solver Code

13. Parallelizing Poisson solver (2) 1. Parallel section begins after ("NTIMES = atoi(argv[3]);") mapping is Linear{ 2. ParExchange2DArrayDouble (a, M, N) before statement ("printMatrix (a, M, N);") && in function ("main"); 3. ParExchange2DArrayDouble (b, M, N) before statement ("printMatrix (a, M, N);") && in function ("main"); 4. ParExchange2DArrayDouble (b, M, N) after statement ("b=compute(a, f, b, M, N);") && in function ("main"); 5. AllReduceSumInt(norm) after statement ("norm = normdiff(b, a, M, N);") && in function ("main") 6. } 13 Hi-PaL Code to Generate MPI Code for Poisson Solver

14. Generated MPI Code for Poisson Solver (1) 1. //other code 2. NTIMES = atoi(argv[3]); 3. MPI_Init(NULL, NULL); 4. MPI_Comm_size(MPI_COMM_WORLD, &size_Fraspa); 5. MPI_Comm_rank(MPI_COMM_WORLD, &rank_Fraspa); 6. create_2dgrid(MPI_COMM_WORLD, &comm2d_Fraspa,…); 7. create_diagcomm(MPI_COMM_WORLD, size_Fraspa, …); 8. rowmap_Fraspa.init(M, P_Fraspa, p_Fraspa); 9. colmap_Fraspa.init(N, Q_Fraspa, q_Fraspa); 10. myrows_Fraspa = rowmap_Fraspa.getMyCount(); 11. mycols_Fraspa = colmap_Fraspa.getMyCount(); 12. M_Fraspa = M; 13. N_Fraspa = N; 14. M = myrows_Fraspa; 15. N = mycols_Fraspa; 16. a = allocMatrix (a, M, N); 14

15. Generated MPI Code for Poisson Solver (2) 17. b = allocMatrix (b, M, N); 18. f = allocMatrix (f, M, N); 19. start = 0; 20. //other code 21. a = exchange (a, myrows_Fraspa + 2, …); 22. b = exchange (b, myrows_Fraspa + 2, …); 23. printMatrix (a, M, N); 24. t1 = MPI_Wtime(); 25. for (k = start; k = tolerance; k++) { 26. b = compute(a, f, b, M, N); 27. b = exchange (b, myrows_Fraspa + 2, …); 28. ptr = a; 29. a = b; 30. b = ptr; 31. norm = normdiff(b, a, M, N); 32. MPI_Allreduce(&norm, &norm_Fraspa, 1, MPI_INT, MPI_SUM,…); 33. norm = norm_Fraspa; 34. } 36. //other code 15

16. Snippet of Exchange Template for MPI template T** exchange(T** data, int nrows, int ncols, int P, int Q, int p, int q, MPI_Comm comm2d, MPI_Comm rowcomm, MPI_Comm colcomm) { //other code above. Create datatype for the recvtype code below MPI_Type_vector(nrows-2, 1, ncols, datatype, &temptype); MPI_Type_extent(datatype, &sizeoftype) ; int blens[2] = {1, 1}; MPI_Aint displ[2] = {0, sizeoftype}; MPI_Datatype types[2] = {temptype, MPI_UB}; MPI_Type_struct (2, blens, displ, types, &vectype); MPI_Type_commit(&vectype); MPI_Cart_shift(rowcomm, 0, -1, &prev, &next); MPI_Cart_shift(colcomm, 0, -1, &down, &up); // send and receive the boundary rows MPI_Irecv(&data[0][1], ncols-2, datatype, up, 0, …); MPI_Irecv(&data[nrows-1][1], ncols-2, datatype, down, 0, …); MPI_Isend(&data[1][1], ncols-2, datatype, up, 0, colcomm, …); MPI_Isend(&data[nrows-2][1], ncols-2, datatype, down, …); …} 16

17. Providing the Specifications Through Command-Line Interface 17 Would you like to use MPI or OpenMP? (1) MPI (2) OpenMP 2 ================================================== Would you like to use MIC? (1) Yes (2) No 1 ================================================== Would you like to use this for loop? Y or N? for ( i = 0; i < ihi; i++ ){ i4_to_bvec ( i, n, bvec ); value = circuit_value (n, bvec);... //other lines of code } Y This loop contains the following variables: i,value,j,solution_num Choose the variables for reduction: solution_num Operation Complete...OpenMP code with offload capability is generated

18. Providing the Specifications Through Graphical User Interface 18

19. Currently Available Support For… Parallel Programming Paradigms MPI OpenMP MPI + OpenMP OpenMP + Offload CUDA Parallel Programming Patterns For-loops with Reduction Stencil-Based Computations (Regular Mesh) PipelineReplicable Base languages supported C/C++ 19 Support for Fortran will be added …

20. Time for demos Using the framework through GUI Using the framework through Hi-PaL 20

21. Outline Motivation and Goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 21

22. Results: Poisson Solver (Hi-PaL based MPI) 22

23. Results: Genetic Algorithm for Content Based Image Retrieval (Hi-PaL based MPI & OpenMP) 23

24. Results: Seismic Tomography Code (GUI-based CUDA) 24

25. 25 Results: Circuit Satisfiability Code (GUI-based OpenMP + Offload)

26. Outline Motivation and Goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 26

27. Summary of Features & Benefits of the Framework Enhances the productivity of the end-users in terms of the reduction in the time and effort – reduction in manual effort by over 90% while ensuring that the performance of the generated parallel code is within 5% of the sample hand-written parallel code Leverages the knowledge of expert parallel programmers Separates the sequential and parallel programming concerns while preserving the existing version of sequential applications 27

28. Outline Motivation and goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 28

29. Project Status 29

30. Outline Motivation and goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 30

31. Future Work Migrate from DMS to Rose source-to-source compiler and integrate all the interfaces Usability studies: – Should be able to rank the user preferences for the interface – Prioritize the development effort Integration with PerfExpert and Eclipse Ability to handle irregular meshes and specify pipeline mode of communication through GUI Address the demands for – a directives-based interface – the option of editing the log-file to repeat the code- generation process without going through the GUI 31

32. Outline Motivation and goals Overview of the Framework with demos Results Features and Benefits Project Status Future Work Conclusion Q & A 32

33. Conclusion Through this research and development effort, we have demonstrated – an approach for lowering the adoption barriers to HPC by raising the level of abstraction of parallel programming – an interactive tool for teaching parallel programming: think “Alice” and “DrJava” – the usage of multiple interfaces to accommodate the preferences of the user-community: one size does not fit all – that it is possible to achieve abstraction and performance at the same time 33

34. Acknowledgement On behalf of my co-authors, student and XSEDE intern (Julio Olaya), I would like to thank NSF, XSEDE and TACC! 34

35. Thanks for listening! Any questions, comments or concerns? Contact: rauta@tacc.utexas.edurauta@tacc.utexas.edu 35