Compiling Parallel Loops for High Performance Computers de David E. Hudak

Compiling Parallel Loops for High Performance Computers: The Springer International Series in Engineering and Computer Science, cartea 200

David E. Hudak

Santosh G. Abraham

4. 2 Code Segments . . . . . . . . . . . . . . . 96 4. 3 Determining Communication Parameters . 99 4. 4 Multicast Communication Overhead · 103 4. 5 Partitioning . . . . . . · 103 4. 6 Experimental Results . 117 4. 7 Conclusion. . . . . . . · 121 5 COLLECTIVE PARTITIONING AND REMAPPING FOR MULTIPLE LOOP NESTS 125 5. 1 Introduction. . . . . . . . . 125 5. 2 Program Enclosure Trees. . 128 5. 3 The CPR Algorithm . . 132 5. 4 Experimental Results. . 141 5. 5 Conclusion. . 146 BIBLIOGRAPHY. 149 INDEX . . . . . . . . 157 LIST OF FIGURES Figure 1. 1 The Butterfly Architecture. . . . . . . . . . 5 1. 2 Example of an iterative data-parallel loop . . 7 1. 3 Contiguous tiling and assignment of an iteration space. 13 2. 1 Communication along a line segment. . . 24 2. 2 Access pattern for the access offset, (3,2). 25 2. 3 Decomposing an access vector along an orthogonal basis set of vectors. . . . . . . . . . . . . . . . . . . 26 2. 4 An analysis of communication patterns. 29 2. 5 Decomposing a vector along two separate basis sets of vectors. 31 2. 6 Cache lines aligning with borders. 33 2. 7 Cache lines not aligned with borders. 34 2. 8 nh is the difference of nd and nb. 42 2. 9 nh is the sum of nd and nb. 42 2. 10 The ADAPT system. 44 2. 11 Code segment used in experiments. . 46 2. 12 Execution rates for various partitions. 47 2. 13 Execution time of partitions on Multimax. 48 2. 14 Performance increase as processing power increases. 49 2. 15 Percentage miss ratios for various aspect ratios and line sizes.

Citește tot Restrânge

Toate formatele și edițiile

	Preț	Express
Paperback (1)	617^.76 lei 6-8 săpt.
Springer – 27 sep 2012	617^.76 lei 6-8 săpt.
Hardback (1)	623^.32 lei 6-8 săpt.
Springer – 31 oct 1992	623^.32 lei 6-8 săpt.

Din seria The Springer International Series in Engineering and Computer Science

18%

Preț: 1177^.92 lei
18%

Preț: 921^.36 lei
20%

Preț: 615^.32 lei
20%

Preț: 621^.64 lei
20%

Preț: 618^.96 lei
24%

Preț: 847^.29 lei
20%

Preț: 622^.37 lei
20%

Preț: 950^.07 lei
20%

Preț: 621^.01 lei
18%

Preț: 903^.90 lei
20%

Preț: 950^.72 lei
18%

Preț: 919^.85 lei
15%

Preț: 621^.23 lei
18%

Preț: 913^.32 lei
18%

Preț: 1173^.85 lei
15%

Preț: 619^.12 lei
18%

Preț: 904^.83 lei
18%

Preț: 904^.83 lei
20%

Preț: 1234^.64 lei
20%

Preț: 1725^.82 lei
20%

Preț: 945^.61 lei
Preț: 376^.90 lei
20%

Preț: 610^.19 lei
18%

Preț: 1193^.58 lei
20%

Preț: 622^.65 lei
20%

Preț: 1234^.18 lei
18%

Preț: 911^.89 lei
20%

Preț: 625^.45 lei
15%

Preț: 617^.25 lei
18%

Preț: 1179^.97 lei
15%

Preț: 624^.94 lei
15%

Preț: 618^.83 lei
20%

Preț: 620^.83 lei
18%

Preț: 923^.62 lei
15%

Preț: 619^.25 lei
18%

Preț: 1180^.20 lei
18%

Preț: 912^.45 lei

Cuprins

1 Introduction.- 1.1 Model Assumptions.- 1.2 Related Work.- 1.3 Overview.- 2 Contiguous Loop Partitions for Neighborhood Communication.- 2.1 Introduction.- 2.2 Quantifying Communication.- 2.3 Compensation for Cache Line Size.- 2.4 Partition Construction.- 2.5 Experimental Evaluation of ADP.- 2.6 Conclusion.- 3 Contiguous Data Assignments for Neighborhood Communication.- 3.1 Introduction.- 3.2 Data Assignments.- 3.3 Exploiting Overlap.- 3.4 Software Redundancy.- 3.5 ADAPT.- 3.6 Experimental Results.- 3.7 Conclusion.- 4 Cyclic Loop Partitions for Linearly Varying Loops.- 4.1 Introduction.- 4.2 Code Segments.- 4.3 Determining Communication Parameters.- 4.4 Multicast Communication Overhead.- 4.5 Partitioning.- 4.6 Experimental Results.- 4.7 Conclusion.- 5 Collective Partitioning and Remapping for Multiple Loop Nests.- 5.1 Introduction.- 5.2 Program Enclosure Trees.- 5.3 The CPR Algorithm.- 5.4 Experimental Results.- 5.5 Conclusion.