SpringerLink Springer Engineering eBooks 2017 English+International

Incluye referencias bibliográficas e índice

Foreword; Preface; Contents; 1 Introduction; 1.1 Sources of Variability; 1.2 Delay Variation; 1.3 Book Organization; References; Part I Predicting and Preventing Errors; 2 Instruction-Level Tolerance; 2.1 Introduction; 2.2 Effect of Operating Conditions; 2.3 Delay Variation Among Pipeline Stages; 2.4 Instruction Characterization Methodology and Experimental Results; 2.4.1 Gate-Level Simulation; 2.4.2 Instruction-Level Delay Variability; 2.4.3 Less Intrusive Variation-Tolerant Technique; 2.4.4 Power Variability; 2.5 Chapter Summary; References; 3 Sequence-Level Tolerance; 3.1 Introduction.

3.2 PVT Variations3.2.1 Conventional Static Timing Analysis; 3.2.2 Variation-Aware Statistical STA; 3.3 Error-Tolerant Applications; 3.3.1 Analysis of Adaptive Guardbanding for Probabilistic Applications; 3.4 Error-Intolerant Applications; 3.4.1 Sequence-Level Vulnerability (SLV); 3.4.2 SLV Characterization; 3.5 Adaptive Guardbanding; 3.6 Experimental Results; 3.6.1 Effectiveness of Adaptive Guardbanding; 3.6.2 Overhead of Adaptive Guardbanding; 3.7 Chapter Summary; References; 4 Procedure-Level Tolerance; 4.1 Introduction; 4.2 Variation-Tolerant Processor Clusters Architecture.

4.2.1 Variation-Aware VDD-Hopping4.3 Procedure Hopping for Dynamic IR-Drop; 4.3.1 Supporting Intra-cluster Procedure Hopping; 4.4 Characterization of PLV to Dynamic Operating Conditions; 4.5 Experimental Results; 4.5.1 Cost of Procedure Hopping; 4.6 Chapter Summary; References; 5 Kernel-Level Tolerance; 5.1 Introduction; 5.2 Device-Level NBTI Model; 5.3 GP-GPU Architecture; 5.3.1 GP-GPU Workload Distribution; 5.4 Aging-Aware Compilation; 5.4.1 Observability: Aging Sensors; 5.4.2 Prediction: Wearout Estimation Module; 5.4.3 Controllability: Uniform Slot Assignment; 5.5 Experimental Results.

5.6 Chapter SummaryReferences; 6 Hierarchically Focused Guardbanding; 6.1 Introduction; 6.2 Timing Error Model for PVTA; 6.2.1 Analysis Flow for Timing Error Extraction; 6.2.2 Parametric Model Fitting; 6.2.3 TER Classification; 6.2.4 Robustness of Classification; 6.3 Runtime Hierarchically Focused Guardbanding; 6.3.1 Observability; 6.3.2 Controllability; 6.4 A Case Study of HFG on GPUs; 6.5 Chapter Summary; References; Part II Detecting and Correcting Errors; 7 Work-Unit Tolerance; 7.1 Introduction; 7.2 Architectural Support for VOMP; 7.3 Work-Unit Vulnerability and VOMP Work-Sharing.

7.3.1 Intra- and Inter-corner WUV7.3.2 Online WUV Characterization; 7.4 VOMP Schedulers; 7.4.1 Variation-Aware Task Scheduling (VATS); 7.4.2 Variation-Aware Section Scheduling (VASS); 7.5 Experimental Results; 7.5.1 Framework Setup; 7.5.2 VOMP Results for Tasking; 7.5.3 VOMP Results for Sections; 7.6 Chapter Summary; References; 8 Memristive-Based Associative Memory for Error Recovery; 8.1 Introduction; 8.2 Energy-Efficient GP-GPUs; 8.2.1 Associative Memristive-Based Computing; 8.3 Collaborative Compilation; 8.3.1 FPU Memristive-Based Computing; 8.4 Experimental Results.

This book focuses on computing devices and their design at various levels to combat variability. The authors provide a review of key concepts with particular emphasis on timing errors caused by various variability sources. They discuss methods to predict and prevent, detect and correct, and finally conditions under which such errors can be accepted; they also consider their implications on cost, performance and quality. Coverage includes a comparative evaluation of methods for deployment across various layers of the system from circuits, architecture, to application software. These can be combined in various ways to achieve specific goals related to observability and controllability of the variability effects, providing means to achieve cross layer or hybrid resilience. · Covers challenges and opportunities in identifying microelectronic variability and the resulting errors at various layers in the system abstraction; · Enables readers to assess how various levels of circuit and system design can mitigate the effects of variability; · Demonstrates overall system architecture of what is now called "approximate computing" paradigm in massively parallel integrated architectures and accelerators.