Recent Trends in Parallel Computing

Recent Trends in Parallel Computing Recent Trends in Parallel Computing [2393-8749(e)] is a peer-reviewed hybrid open-access journal launched in 2014. Parallel computing is a form of computation in which many calculations can be done at the same time and it works on the principle that large problems can often be divided into smaller ones, which are then solved in parallel. Specialized parallel computer architectures are sometimes used aboard traditional processors, to quicken specific tasks this increases the speed of execution of the task.

• Parallel and multicore architectures: shared and distributed memory systems, chip multiprocessors and manycore designs, homogeneous and heterogeneous multicore organisation, instruction and thread-level parallelism, vector and SIMD extensions, and architectural simulation and evaluation.
• Interconnection networks and on-chip communication: network topology design and analysis, mesh, torus, hypercube, and hierarchical topologies, network-on-chip architecture and routing, deterministic and adaptive routing protocols, collective communication patterns, bandwidth and latency characterisation, and scalability of interconnect fabrics.
• Memory systems and coherence: cache hierarchy design for parallel systems, coherence protocols and directory schemes, memory consistency models, non-uniform memory access, high-bandwidth and persistent memory, data placement and locality optimisation, and memory bandwidth bottleneck analysis.
• Parallel programming models and languages: message passing and MPI, shared memory models including OpenMP and Pthreads, partitioned global address space languages, task-based and dataflow models, algorithmic skeletons and parallel patterns, functional and declarative parallelism, and productivity and portability of parallel code.
• GPU and accelerator computing: GPU architecture and programming models, kernel design and optimisation, multi-GPU and multi-node scaling, FPGA and domain-specific accelerators, hybrid CPU–GPU algorithms, accelerated numerical libraries, and performance portability across accelerator platforms.
• Heterogeneous and hybrid computing: workload partitioning across heterogeneous resources, dynamic core and device assignment, runtime systems for heterogeneous platforms, programming abstractions for mixed architectures, and evaluation of heterogeneous versus homogeneous designs.
• Parallel algorithms and numerical methods: design and analysis of parallel algorithms, dense and sparse linear algebra, factorisation methods and iterative solvers, parallel graph algorithms, mixed-precision computation, parallel discrete event simulation, and communication-avoiding algorithm design.
• Scheduling, load balancing, and resource management: static and dynamic scheduling policies, task mapping and placement, load imbalance detection and correction, work stealing and dynamic partitioning, job scheduling on shared clusters, and resource allocation under contention.
• Performance modelling and analysis: speedup and scalability analysis, cost and execution models, profiling and tracing tools, bottleneck identification, benchmark design and workload characterisation, and performance prediction for large-scale systems.
• High-performance and exascale computing: supercomputer system design and deployment, scalable runtime and system software, I/O and parallel file systems, checkpointing and resilience at scale, workflow management for HPC, and convergence of HPC with data-intensive computing.
• Distributed, cloud, and edge parallelism: distributed computing frameworks, cloud-based and elastic parallel execution, serverless and container-based parallel workloads, grid and volunteer computing, and parallelism across edge and fog resources.
• Fault tolerance and reliability: fault-tolerant interconnect and routing, degraded-mode operation and graceful degradation, error detection, recovery, and rollback, reliability modelling of large parallel systems, and resilience of long-running computations.
• Energy, power, and thermal management: power-aware scheduling and voltage and frequency scaling, thermal management and activity migration, energy–performance trade-offs, cooling and infrastructure considerations, and energy efficiency metrics for parallel systems.
• Applications of parallel computing: scientific and engineering simulation, computational fluid dynamics and finite element analysis, distributed training of machine learning models, bioinformatics and phylogenetics, image and signal processing workloads, and large-scale data analytics.