Cargando…
Advances in GPU research and practice /
Advances in GPU Research and Practice focuses on research and practices in GPU based systems. The topics treated cover a range of issues, ranging from hardware and architectural issues, to high level issues, such as application systems, parallel programming, middleware, and power and energy issues....
| Clasificación: | Libro Electrónico |
|---|---|
| Otros Autores: | |
| Formato: | Electrónico eBook |
| Idioma: | Inglés |
| Publicado: |
Amsterdam :
Elsevier,
[2017]
|
| Colección: | Emerging trends in computer science & applied computing.
|
| Temas: | |
| Acceso en línea: | Texto completo (Requiere registro previo con correo institucional) |
Tabla de Contenidos:
- Front Cover; Advances in GPU Research and Practice; Copyright; Dedication; Contents; List of Contributors; Preface; Acknowledgments; Part 1: Programming and tools; Chapter 1: Formal analysis techniques for reliable GPU programming: current solutions and call to action; 1 GPUs in Support of Parallel Computing; Bugs in parallel and GPU code; 2 A quick introduction to GPUs; Organization of threads; Memory spaces; Barrier synchronization; Warps and lock-step execution; Dot product example; 3 Correctness issues in GPU programming; Data races; Lack of forward progress guarantees.
- Floating-point accuracy4 The need for effective tools; 4.1 A Taxonomy of Current Tools; 4.2 Canonical Schedules and the Two-Thread Reduction; Race freedom implies determinism; Detecting races: ``all for one and one for all''; Restricting to a canonical schedule; Reduction to a pair of threads; 4.3 Symbolic Bug-Finding Case Study: GKLEE; 4.4 Verification Case Study: GPUVerify; 5 Call to Action; GPUs will become more pervasive; Current tools show promise; Solving basic correctness issues; Equivalence checking; Clarity from vendors and standards bodies; User validation of tools; Acknowledgments.
- 4.5 Detecting Memory Objects Written by a Kernel5 SnuCL extensions to OpenCL; 6 Performance evaluation; 6.1 Evaluation Methodology; 6.2 Performance; 6.2.1 Scalability on the medium-scale GPU cluster; 6.2.2 Scalability on the large-scale CPU cluster; 7 Conclusions; Acknowledgments; References; Chapter 3: Thread communication and synchronization on massively parallel GPUs; 1 Introduction; 2 Coarse-Grained Communication and Synchronization; 2.1 Global Barrier at the Kernel Level; 2.2 Local Barrier at the Work-Group Level; 2.3 Implicit Barrier at the Wavefront Level.
- 3 Built-In Atomic Functions on Regular Variables4 Fine-Grained Communication and Synchronization; 4.1 Memory Consistency Model; 4.1.1 Sequential consistency; 4.1.2 Relaxed consistency; 4.2 The OpenCL 2.0 Memory Model; 4.2.1 Relationships between two memory operations; 4.2.2 Special atomic operations and stand-alone memory fence; 4.2.3 Release and acquire semantics; 4.2.4 Memory order parameters; 4.2.5 Memory scope parameters; 5 Conclusion and Future Research Direction; References; Chapter 4: Software-level task scheduling on GPUs; 1 Introduction, Problem Statement, and Context.