عنوان

Tools for high performance computing 2017 :

3030119866

3030119874

3030119882

3030119890

9783030119867

9783030119874

9783030119881

9783030119898

9783030119867

Tools for high performance computing 2017 :

[Book]

proceedings of the 11th International Workshop on Parallel Tools for High Performance Computing, September 2017, Dresden, Germany /

editors, Christoph Niethammer, Michael M. Resch, Wolfgang E. Nagel, Holger Brunst and Hartmut Mix.

Cham, Switzerland :

Springer,

2019.

©2019

1 online resource :

color illustrations

Includes bibliographical references.

Intro; Preface; Contents; A Structured Approach to Performance Analysis; 1 Introduction; 2 Background; 2.1 The EU Centre of Excellence for Performance Optimization and Productivity (POP); 2.2 The BSC Tools; 2.3 Example Code: CheSS; 3 Structured Performance Analysis; 3.1 Measurement; 3.2 Overview and Focus of Analysis; 3.3 Performance Modeling; 3.4 Detailed Analysis; 3.5 Reporting; 4 Conclusions; References; Counter Inspection Toolkit: Making Sense Out of Hardware Performance Events; 1 Introduction; 2 Non-obvious Code Behavior; 3 Branch-Related Events; 3.1 Design Choices

3.2 Controlling Branch Misprediction3.3 Event Categories; 3.4 Analysis of Benchmark Results; 4 Cache-Related Events; 4.1 Assisting Developers with Code Optimization; 5 Categorizing Events Automatically; 6 Related Work; 7 Conclusions; References; ASSIST: An FDO Source-to-Source Transformation Tool for HPC Applications; 1 Introduction; 2 Background and Goals; 3 Design and Implementation; 3.1 Overview; 3.2 Compiler Infrastructure; 3.3 Integration into MAQAO; 4 Supported Transformations; 4.1 Common Loop Transformations; 4.2 Constant Propagation and Local Dead Code Elimination; 4.3 Specialization

4 Vectorization Assistance4.1 Metrics; 5 Measurement Work-Flow; 5.1 Multi Run; 5.2 Cube Tools; 6 Topology Visualization; 7 Conclusion; References; Towards Elastic Resource Management; 1 Introduction; 2 Theoretical Background on Multiprocessor Scheduling; 2.1 Problem Statement; 2.2 Computational Complexity; 2.3 Resource-Static Scheduling in Distributed Memory HPC Systems; 2.4 Modified Scheduling Problem for Resource-Elastic Execution; 3 Performance Monitoring Infrastructure; 3.1 Process-Local Pattern Detection and Performance Measurements; 3.2 Node-Local Reductions and Performance Data Updates

4.4 Loop Count Transformation4.5 Block Vectorization Transformation; 5 Experiments; 6 Related Work; 7 Conclusion and Future Work; References; Unifying the Analysis of Performance Event Streams at the Consumer Interface Level; 1 Introduction; 2 Components in the Instrumentation Chain; 2.1 General Model; 2.2 Event Sources; 2.3 Intermediate Representations; 2.4 Event Forwarding; 2.5 Event Analysis; 2.6 Tools Interoperability; 3 Towards a Shared-Representation of Performance Events; 3.1 Towards a Tool Network; 3.2 The Producer/Consumer Interface; 4 Practical Illustration with MALP; 5 Conclusion

6 Future WorkReferences; OMPT-Multiplex: Nesting of OMPT Tools; 1 Introduction; 2 Use Cases; 2.1 Affinity Display; 2.2 Supplementary Debugging Tool; 2.3 Other Tools Building on OMPT; 3 OMPT Multiplex Architecture; 4 Activating an OMPT Tool; 5 Tool Data Pointer; 6 Initializing an OMPT Tool; 7 Runtime Entry Points; 8 Callback Functions; 9 OMPT-Multiplex Versus OMPTn; 10 Recursive Use of OMPT Multiplex; 11 Conclusion; References; SCIPHI Score-P and Cube Extensions for Intel Phi; 1 Introduction; 2 Related Work; 3 Memory Analysis; 3.1 DDR4 SDRAM and MCDRAM Usage; 3.2 MCDRAM Candidates

0

8

8

8

8

This book presents the proceedings of the 11th International Parallel Tools Workshop, a forum to discuss the latest advances in parallel tools, held September 11-12, 2017 in Dresden, Germany. High-performance computing plays an increasingly important role for numerical simulation and modeling in academic and industrial research. At the same time, using large-scale parallel systems efficiently is becoming more difficult. A number of tools addressing parallel program development and analysis has emerged from the high-performance computing community over the last decade, and what may have started as a collection of a small helper scripts has now matured into production-grade frameworks. Powerful user interfaces and an extensive body of documentation together create a user-friendly environment for parallel tools.

Springer Nature

com.springer.onix.9783030119874

9783030119867

9783030119881

9783030119898

High performance computing, Congresses.

High performance computing.

MATHEMATICS-- Essays.

MATHEMATICS-- Pre-Calculus.

MATHEMATICS-- Reference.

MAT-- 023000

MAT-- 026000

MAT-- 039000

PDE

PDE

Brunst, Holger.

Mix, Hartmut

Nagel, Wolfgang E.

Niethammer, Christoph

Resch, Michael M.

International Workshop on Parallel Tools for High Performance Computing(11th :2017 :, Dresden, Germany)

20200823082800.0

pn

[Book]

Y