Parallel and Distributed Computing, Applications and Technologies: 23rd International Conference, PDCAT 2022, Sendai, Japan, December 7–9, 2022, Proceedings

✍ Scribed by Hiroyuki Takizawa, Hong Shen, Toshihiro Hanawa, Jong Hyuk Park, Hui Tian, Ryusuke Egawa

• Publisher: Springer
• Year: 2023
• Tongue: English
• Leaves: 526
• Series: Lecture Notes in Computer Science, 13798
• Category: Library

✦ Synopsis

This book constitutes the proceedings of the 23rd International Conference on Parallel and Distributed Computing, Applications, and Technologies, PDCAT 2022, which took place in Sendai, Japan, during December 7-9, 2022.

The 24 full papers and 16 short papers included in this volume were carefully reviewed and selected from 95 submissions. The papers are categorized into the following topical sub-headings: Heterogeneous System (1; HPC & AI; Embedded systems & Communication; Blockchain; Deep Learning; Quantum Computing & Programming Language; Best Papers; Heterogeneous System (2); Equivalence Checking & Model checking; Interconnect; Optimization (1); Optimization (2); Privacy; and Workflow.

✦ Table of Contents

Preface
Organization
Contents
Heterogeneous System (1)
Towards Priority-Flexible Task Mapping for Heterogeneous Multi-core NUMA Systems
1 Introduction
2 Motivation and Objective
3 Task Mapping with Priority Option Switching
3.1 Overview
3.2 Potential Benefit Metrics and POSM
3.3 HPO and MPO Mapping Algorithms
4 Evaluation
4.1 Experimental Environment
4.2 Evaluation of HPO and MPO
4.3 Evaluation of POSM
5 Conclusions
References
Multi-GPU Scaling of a Conservative Weakly Compressible Solver for Large-Scale Two-Phase Flow Simulation
1 Introduction
2 Numerical Methods
2.1 Con-CAC-LS
2.2 Con-PLIC-HF
2.3 Evolving Pressure Projection Method
3 Multi-GPU Computation
3.1 Performance of Con-CAC-LS
3.2 Performance of Con-PLIC-HF
4 Numerical Results
4.1 Rayleigh-Taylor Instability
4.2 Drop Impacting on a Thin Liquid Film
4.3 Liquid Jet in Gas Cross-Flow
5 Conclusion
References
Improving the Performance of Lattice Boltzmann Method with Pipelined Algorithm on A Heterogeneous Multi-zone Processor
1 Introduction
2 MT-3000
2.1 Programming Environment
3 Methodology
3.1 Lattice Boltzmann Method
3.2 Data Storage Schemes
3.3 An Improved Pipelined Algorithm
3.4 Multi-level Parallelization Strategy
4 Numerical and Performance Results
5 Conclusions
References
FPGA
DEEPFAKE CLI: Accelerated Deepfake Detection Using FPGAs
1 Introduction
2 Motivation and Background
3 Implementation
3.1 Software Implementation
3.2 Hardware Implementation
4 Results
4.1 Software Results
4.2 Hardware Results
4.3 Benchmarking Deepfake CLI with Other Nodes
5 Conclusion
6 Future Work
References
Memory Access Optimization for Former Process of Pencil Drawing Style Image Conversion in High-Level Synthesis
1 Introduction
2 Pencil Drawing Style Conversion Process
3 Proposal Method
3.1 Reconstruction of Edge Strength Image Input/output
3.2 Reconstruction of Memory Access Unit
4 Experiments and Discussions
4.1 Software Execution Time
4.2 Simplified Estimate of the Number of Cycles After High-Level Synthesis
4.3 Performance Estimation in Logic Circuit Simulation
4.4 Performance Measurement on Actual Equipment
5 Conclusion
References
Word2Vec FPGA Accelerator Based on Spatial and Temporal Parallelism
1 Introduction
2 Previous Work
2.1 Word2vec Algorithm
2.2 Acceleration of Word2vec
2.3 Data-Flow of Word2vec
3 Systolic Array Architecture for Word2vec
4 Evaluation
5 Conclusion
References
HPC and AI
Analyzing I/O Performance of a Hierarchical HPC Storage System for Distributed Deep Learning
1 Introduction
2 Background
2.1 File Access in Distributed Neural Network Workloads
2.2 Storage System in HPC
3 Related Work
4 Methodology
4.1 Overview
4.2 Measuring I/O Performance by Benchmark
4.3 Analyzing the I/O Performance
4.4 Estimate Performance by Storage Improvement
5 Experiment Results
5.1 Setup for Experiment
5.2 Measuring Execution Time for Epochs
5.3 Analyzing I/O Performance
5.4 Estimating the Impact of the Storage Improvements
6 Discussion
7 Conclusion
References
An Advantage Actor-Critic Deep Reinforcement Learning Method for Power Management in HPC Systems
1 Introduction
2 Power State Management Problem
3 A2C Deep Reinforcement Learning for Power State Management Problem
3.1 MDP Formulation of PSMP
3.2 A2C-DRL for PSMP
4 Evaluation
4.1 Experimental Setup
4.2 Results and Discussion
5 Conclusion
References
An AutoML Based Algorithm for Performance Prediction in HPC Systems
1 Introduction
2 Algorithm
3 SPEC Benchmark Applications Performance Dataset
4 Experiments and Results
4.1 Evaluation of Algorithm Effectiveness
4.2 Accuracy Comparison with the State-of-the-Art
5 Conclusions and Future Work
References
Embedded Systems and Communication
Edge-Gateway Intrusion Detection for Smart Home
1 Introduction
2 Related Work
3 Edge-Gateway Intrusion Detection System for Smart Home
3.1 System Framework
3.2 Edge Intrusion Detection of Smart Device Based on Gaussian Distribution
3.3 Gateway-Level Centralized Intrusion Detection Based on RF-GraphSAGE
4 Results and Discussion
4.1 Experimental Setup
4.2 Results of Device-Level Edge Intrusion Detection
4.3 Results of Gateway-Level Centralized Intrusion Detection
5 Conclusion
References
Energy-Delay Tradeoff in Parallel Task Allocation and Execution for Autonomous Platooning Applications
1 Introduction
2 Divisible Load Scheduling Model
3 Markov Chain Model
4 Energy Consumption Model
4.1 Energy Consumption Model for Task Execution
4.2 Energy Consumption Model for Load Transmission
5 Energy Consumption-Delay Tradeoff Task Allocation and Execution Algorithm
6 Simulation and Analysis
7 Conclusion
References
A Reservation-Based List Scheduling for Embedded Systems with Memory Constraints
1 Introduction
2 Related Work
2.1 Scheduling with Shared Memory Constraints
2.2 Analysis
3 Challenges: Scheduling with Memory Constraints
3.1 Memory Constraints
3.2 Motivation Example
4 Problem Modeling
4.1 Memory Model
4.2 List Scheduling Definition
5 Proposed Solution
6 Experimental Results
6.1 Random Workflows Generator
6.2 Impact of Depth k
6.3 Minimum Memory Usage
6.4 Minimized Makespan Value Under Memory Constraints
6.5 Applicability
7 Conclusions
References
Formalization and Verification of SIP Using CSP
1 Introduction
2 Background
2.1 SIP
2.2 SDN
3 Modeling SIP
3.1 Sets, Messages and Channels
3.2 Overall Modeling
3.3 Register
3.4 Client
3.5 Intruder
4 Verification
4.1 Verification in PAT
4.2 Results
5 Improvement
5.1 Modeling SIPS
5.2 Verification
6 Conclusion and Future Work
References
Blockchain
Towards a Blockchain and Fog-Based Proactive Data Distribution Framework for ICN
1 Introduction
2 Related Work
3 Proposed Framework
3.1 Framework Overview
3.2 Framework Data Flow
3.3 Security-Related Advantages
3.4 Block Structure
4 Performance Results
4.1 Lesson Learned
5 Conclusion and Future Work
References
Research on User Influence Weighted Scoring Algorithm Incorporating Incentive Mechanism
1 Introduction
2 User Influence Weighting Algorithm
2.1 User Influence Model
2.2 Weighted Rating Algorithm
3 Rating Incentive Mechanism
4 Experiments and Analysis
4.1 Experimental Environment
4.2 Experimental Protocols
4.3 Analysis of Experimental Results
4.4 Safety Analysis
5 Conclusion
References
BloodMan-Chain: A Management of Blood and Its Products Transportation Based on Blockchain Approach
1 Introduction
2 Related Work
2.1 Blood Supply Chain Management Systems Not Based on Blockchain Technology
2.2 Blood Supply Chain Management Systems Based on Blockchain Technology
3 BloodMan-Chain Architecture
3.1 Overview Model
3.2 Detailed Model
4 Evaluation Scenarios
4.1 Environment Setting
4.2 Results
5 Conclusion
References
Deep Learning
A Systematic Comparison on Prevailing Intrusion Detection Models
1 Introduction
2 Related Work
3 Methodology
4 Experimental Study
5 Conclusions
References
Enhancing Resolution of Inferring Hi-C Data Integrating U-Net and ResNet Networks
1 Introduction
2 Method
2.1 Data Preprocessing
2.2 Structure of Network Model
2.3 Algorithm
3 Experiment
3.1 Experiment of Training Network Model
3.2 Experimental Results on Human Cell Datasets
4 Conclusion
References
Detecting Network Intrusions with Resilient Approaches Based on Convolutional Neural Networks
1 Introduction
2 Deep Learning Applications in Network Intrusion Detection
3 Application of CNN-LSTM on NSL-KDD Dataset
3.1 NSL-KDD Dataset Overview
3.2 Proposed CNN-LSTM Model
4 Experiment and Results
4.1 CNN Model Experiment
4.2 CNN-LSTM Model Experiment
4.3 Models’ Results Comparison
5 Conclusion
References
Quantum Computing and Programming Language
Analysis of Precision Vectors for Ising-Based Linear Regression
1 Introduction
2 Ising-Based Linear Regression
2.1 Linear Regression
2.2 A Precision Vector
2.3 A Precision Vector of Ising-Based Linear Regression
3 Analysis of a Precision Vector for Efficient Ising-Based Linear Regression
4 Evaluation
4.1 Experimental Environments
4.2 Results and Discussions
5 Related Work
6 Conclusions
References
Evaluating and Analyzing Irregular Tree Search in the Tascell and HOPE Parallel Programming Languages
1 Introduction
2 Irregular Tree Search
3 Conventional Language
4 Tascell and HOPE Languages
5 Evaluations
6 Conclusion
References
Best Papers
Distributed Parallel Tall-Skinny QR Factorization: Performance Evaluation of Various Algorithms on Various Systems
1 Introduction
2 Problem Setting and Overview of Algorithms
2.1 Problem Setting
2.2 Overview of Algorithms
3 Summary of the Target Algorithms
3.1 HQR: Householder QR
3.2 CGS2: Classical Gram-Schmidt with Reorthogonalization
3.3 TSQR
3.4 S-CholQR3: Shifted CholekyQR3
4 Performance Evaluation
4.1 Overview of the Implementation of the Algorithms
4.2 Computational Environments and Evaluation Settings
4.3 Results
5 Related Work
6 Conclusion
References
A Partitioned Memory Architecture with Prefetching for Efficient Video Encoders
1 Introduction
2 Memory Subsystems for Video Encoding
2.1 Video Encoding
2.2 Hardware Encoding Pipeline
2.3 Analysis of Memory Access Pattern
3 A Partitioned Memory System for CTU-Pipelined Video Encoders
3.1 Split Cache Structure
3.2 Coding Tree Unit Prefetcher
4 Evaluations and Discussions
4.1 Experimental Methodology
4.2 Evaluation Results
5 Related Work
6 Conclusions
References
A Hardware Trojan Exploiting Coherence Protocol on NoCs
1 Introduction
2 Related Work
2.1 Hardware Trojan Embedded in Network Interfaces
2.2 Hardware Trojan Colluding with Malware
2.3 Attack by Hardware Trojan Using Messages Controlling Data Coherence
2.4 Packet Tampering Attack
3 New Security Risk
3.1 Prerequisite
3.2 New Eavesdropping Attack
3.3 Denial-of-Service Attack
4 Countermeasure
4.1 Detection Method
4.2 Countermeasure
5 Evaluation
5.1 Execution Time and Traffic Amount
5.2 Energy Consumption
5.3 Hardware Amount
6 Conclusions
References
A System-Wide Communication to Couple Multiple MPI Programs for Heterogeneous Computing
1 Introduction
2 Heterogeneous Coupling Computing as a Next Computing Environment
3 Heterogeneous Coupling Communication Requirements
4 WaitIO-Socket Design
5 WaitIO-Socket Implemanetation
5.1 WaitIO Overview and Application Program Interface
5.2 Implementation Overview
6 WaitIO-Socket Evaluation
6.1 Evaluation Environment
6.2 Multiple Stream Performance
6.3 Wisteria System Performance
6.4 Application Performance
7 Related Work
8 Summary
References
Heterogeneous System (2)
A Task-Parallel Runtime for Heterogeneous Multi-node Vector Systems
1 Introduction
2 Related Work
2.1 Task-Parallel Runtime System
2.2 SX-Aurora TSUBASA
3 A Task-Parallel Runtime for SX-AT
3.1 An Overview of the Proposed Runtime
3.2 Hybrid MPI for SX-Aurora TSUBASA
4 Evaluation and Discussions
4.1 Evaluation Setup
4.2 Performance Gain by Proposed Runtime
5 Conclusion
References
Accelerating Radiative Transfer Simulation on NVIDIA GPUs with OpenACC
1 Introduction
2 ARGOT: Radiative Transfer Simulation Code for Astrophysics
2.1 ARGOT Method
2.2 ART Method
3 GPU Implementation with OpenACC
3.1 OpenACC
3.2 OpenACC Implementation of the ARGOT Code
4 OpenACC Implementation Using Multiple GPUs
4.1 Overview
4.2 Node Parallelization for ARGOT Code
5 Evaluation
5.1 Experimental Settings
5.2 Performance Comparison Between CUDA and OpenACC Implementations
6 Related Work
7 Conclusion
References
QR Factorization of Block Low-Rank Matrices on Multi-instance GPU
1 Introduction
2 Related Work
3 Numerical Calculations on MIG
3.1 Small Numerical Calculations on GPU
3.2 Small Numerical Calculations on MIG
4 BLR-QR on GPU and MIG
4.1 BLR-QR
4.2 BLR-QR on GPU
4.3 BLR-QR on MIG
5 Performance Evaluation
5.1 Execution Environment and Target Matrices
5.2 Performance Evaluation
6 Conclusions and Future Work
References
Equivalence Checking and Model Checking
Equivalence Checking of Code Transformation by Numerical and Symbolic Approaches
1 Introduction
2 Related Work
2.1 Xevolver for C
2.2 CIVL
3 Equivalence Checking Method
3.1 Extension of Xev-C Notation for Symbolic Execution
3.2 Checking by Numerical Comparison
4 Evaluation
4.1 Evaluation Setup
4.2 Evaluation Results
5 Concluding Remarks and Future Work
References
MEA: A Framework for Model Checking of Mutual Exclusion Algorithms Focusing on Atomicity
1 Introduction
2 Background
2.1 The Mutual Exclusion Algorithms
2.2 Maude
3 The Methodology
3.1 The Snapshot and Transition Rules
3.2 Transition Rules
3.3 Verification
4 Case Study
4.1 High-Level Abstraction
4.2 Refinement
5 Conclusion
References
Interconnect
A High-Radix Circulant Network Topology for Efficient Collective Communication
1 Introduction
2 Background
2.1 Circulant Network Topology
2.2 Target Collective Communication
3 Using Circulant Network Topology
3.1 Definition of Circulant
3.2 Collective-Communication Operations on Circulant Network Topology
4 Evaluation
4.1 Methodology
4.2 Diameter and ASPL
4.3 Hop Counts of Collective-Communication Operations
4.4 Execution Time of Collective-Communication Operations
5 Discussions
6 Conclusions
References
Fault Tolerance and Packet Latency of Peer Fat-Trees
1 Introduction
2 Related Fat-Tree Networks
2.1 k-ary n-fly Butterfly Network
2.2 k-ary n-tree Clos Network
2.3 k-ary n-tree Fat-Tree Network
2.4 Bidirectional k-ary n-tree Clos Network
2.5 k-ary Fat-Tree Network
2.6 Mirrored k-ary n-tree Network
3 Peer k-ary n-tree Network
4 Routing in Peer k-ary n-tree Network
5 Fault Tolerance of Peer k-ary n-tree Network
6 Packet Latency of Peer k-ary n-tree Network
7 Conclusions
References
Accelerating Imbalanced Many-to-Many Communication with Systematic Delay Insertion
1 Introduction
2 Related Work
3 Preliminaries
3.1 Communication Structure
3.2 Communication Schedule
4 Proposed Method
4.1 Arrival-Consumption Model
4.2 Model Parameter Estimation
4.3 Scheduling Algorithm
5 Evaluation
6 Conclusion
References
Optimization (1)
Optimizing Depthwise Convolutions on ARMv8 Architecture
1 Introduction
2 Analysis of Existing Implementations
3 Our Approach
3.1 Implementation
3.2 Arithmetic Intensity
4 Experimental Evaluation
4.1 Experimental Setup
4.2 Algorithm Performance
4.3 Full Topology Performance
5 Conclusion
References
A Profiling-Based Approach to Cache Partitioning of Program Data
1 Introduction
2 Background
2.1 Reuse Distance
2.2 Cache Partitioning
3 Implementation and Experimental Methodology
3.1 Profiler Components
3.2 Cache Partitioning Policy
3.3 Limitations
4 Experimental Results
4.1 Experimental Setup
4.2 Workload
4.3 Dense Matrix Transposed Vector Multiplication
4.4 NAS Parallel Benchmarks
5 Related Work
6 Discussion and Conclusions
References
Optimization (2)
Memory Bandwidth Conservation for SpMV Kernels Through Adaptive Lossy Data Compression
1 Introduction
2 Background
2.1 SpMV Kernel Overview
2.2 Performance Characteristics in SpMV Kernel
2.3 Related Work
3 In-Memory Data Compression Architecture
3.1 Data Compression Memory Interface
3.2 Data Compression Scheme
4 Evaluation
4.1 Methodology
4.2 Empirical Compression Ratio
4.3 Impact on Convergence Conditions by Lower Precision Data
4.4 Cycle-Accurate Performance Evaluation
5 Conclusions
References
SimdFSM: An Adaptive Vectorization of Finite State Machines for Speculative Execution
1 Introduction
2 Background
2.1 Speculative Execution — spec(K)
2.2 Vector Extension and Processor Microarchitecture
2.3 FSM Size — The Number of States |Q|
3 Vectorized FSMs
3.1 Gather
3.2 Shuffle
3.3 Permute
3.4 Interleaved-Gather (iGather)
4 The Adaptive Strategy of SimdFSM
5 Evaluation
5.1 Performance Profiling of Speculating Phase
5.2 Evaluation of the Adaptive Vectorization Strategy
6 Related Work
7 Conclusion
References
Privacy
Broad Learning Inference Based on Fully Homomorphic Encryption
1 Introduction
2 Related Work
2.1 Homomorphic Encryption
2.2 Leveled Homomorphic Encryption Inference
2.3 Privacy Preserving Broad Learning
3 Our Method
3.1 LHE-Based Broad Learning System
3.2 HE-Based Broad Learning System with Incremental Learning
3.3 Polynomial Approximation of Activation Function
4 Experimental Evaluation
4.1 Datasets
4.2 Network Architecture
4.3 Evaluation Method
4.4 Evaluation Results
5 Conclusions
References
Application of Probabilistic Common Set on an Open World Set for Vertical Federated Learning
1 Introduction
2 Background
2.1 Collaboration Between Different Industries and Privacy Care
2.2 Federated Learning
3 Proposed Method
3.1 Basic Idea
3.2 Procedure
4 Evaluation
4.1 Evaluation Environment
4.2 Results and Discussion
5 Conclusion
References
Workflow
Towards a Standard Process Management Infrastructure for Workflows Using Python
1 Introduction
2 Background
3 Approach
3.1 PMIx Python Bindings
3.2 Pyrun Prototype
4 Evaluation/Demonstration
4.1 Single Core Task Injection
4.2 Comparison with Command Line Tools
4.3 Multithreaded Launch Capability
4.4 Heterogeneous Parallel MPI Tasks
5 Related Work
6 Conclusion
References
Author Index