Static and dynamic locality optimizations using integer linear programming

IEEE Transactions on Parallel and Distributed Systems

Volumn 12, Issue 9, 2001, Pages 922-941

  Kandemir, Mahmut ^a,b   Banerjee, Prithviraj ^a,c   Choudhary, Alok ^a,c   Ramanujam, J ^a,d   Ayguadé, Eduard ^e  

^a IEEE   (United States)

^b The Pennsylvania State University   (United States)

^c Northwestern University   (United States)

^d LOUISIANA STATE UNIVERSITY   (United States)

^e European Center for Parallelism of Barcelona   (Spain)

Author keywords

Cache locality; Cache miss estimation; Compiler optimizations; Data reuse; Integer linear programming; Memory layouts

Indexed keywords

DATA REUSE; INTEGER LINEAR PROGRAMMING (ILP);

ALGORITHMS; CACHE MEMORY; DATA REDUCTION; DATA STRUCTURES; HEURISTIC METHODS; INTEGER PROGRAMMING; LINEAR PROGRAMMING; OPTIMIZATION;

PROGRAM COMPILERS;

EID: 0035439109     PISSN: 10459219     EISSN: None     Source Type: Journal    
DOI: 10.1109/TPDS.2001.1184186     Document Type: Article

References (61)

1. Data and computation transformations for multiprocessors
2. Partitioning the statement per iteration space using non-singular matrices
3. Combining optimization for cache and instruction-level parallelism
4. Data-distribution support on distributed-shared memory multiprocessors
5. Application performance on the MIT alewife machine
6. Unifying data and control transformations for distributed shared memory machines
7. Tile size selection using cache organization and data layout
8. New CPU benchmark suites from SPEC
9. Eliminating false sharing
10. On estimating and enhancing cache effectiveness
11. Strategies for cache and local memory management by global program transformations
12. A novel approach towards automatic data distribution
13. Dynamic data distribution with control flow analysis
14. Demonstration of automatic data partitioning techniques for parallelizing compilers on multicomputers
15. Reducing false sharing on shared memory multiprocessors through compile time data transformations
16. Reduction of cache coherence overhead by compiler data layout and loop transformation
17. An integer linear programming approach for optimizing cache locality
18. A hyperplane based approach for optimizing spatial locality in loop nests
19. A matrix-based approach to the global locality optimization problem
20. ImprovingLocality using loop and data transformations in an integrated approach
21. A graph based framework to detect optimal memory layouts for improving data locality
22. Locality optimization algorithms for compilation of out-of-core codes
23. A compiler algorithm for optimizing locality in loop nests
24. Automatic data layout for high performance fortran
25. Transformations of imperfectly nested loops
26. Data-centric multi-level blocking
27. The cache performance and optimizations of blocked algorithms
28. Compiling communication efficient programs for massively parallel machines
29. Improving data locality with loop transformations
30. A quantitive analysis of loop nest locality
31. Non-singular data transformations: Definition, validity, applications
32. Integrating loop and data transformations for global optimisation
33. Automatic selection of dynamic data partitioning schemes for distributed-memory multicomputers
34. The perfect club benchmarks: Effective performance evaluation of supercomputers
35. Parafrase-2: An environment for parallelizing, partitioning, synchronizing, and scheduling programs on multiprocessors
36. Data transformations for eliminating conflict misses
37. The combined effectiveness of unimodular transformations, tiling, and software prefetching
38. Locality analysis for distributed ahared-memory multiprocessors
39. Automatic partitioning of data and computations on scalable shared memory multiprocessors
40. Impact of cache interferences on usual numerical dense loop nests
41. False sharing and spatial locality in multiprocessor caches
42. A data locality optimizing algorithm
43. Combining loop transformations considering caches and scheduling
44. More iteration space tiling
45. Performance analysis using the MIPS R10000 performance counters