Mid_1_PDC_topics.md

• OS Concepts
  • Process vs Threads
  • Serial vs Parallel Execution
  • Singlge vs Multithreading Process
• Parallel Computing
  • Computational Problem
  • Applications
  • Why?
  • Limitations Of Serial Computing
  • Amdahl's Law
• Von Neumann Architecture
  • Direct Memory Access ( DMA )
  • Flynn's Taxonomy
    • SISD
    • SIMD
    • MISD
    • MIMD
• Parallel Terminology
  • Task
  • Parallel Task
  • Serial Execution
  • Parallel Execution
  • Shared Memory
  • Distributed Memory
  • Communication
  • Synchronization
  • Granularity
    • Fine grained
    • Coarse grained
  • Observed Speedup
  • Parallel Overhead
  • Massively Parllel
  • Scalability
• Scope Of Parallelism
  • Implicit Parallelism: Trends in Microprocessor Architectures
  • Pipelining And Superscalar Execution
    • Performance
    • Limitations
    • True data dependency
    • Resource dependency
    • Branch dependency
    • ISsue mechanisms
    • Efficiency Considerations
  • Very Long Instruction Word (VLIW) Processors
    • Considerations
    • Limitations of memory system performance
    • Bandwidth and latency
    • Memory latency
    • Latency using caches
    • Impact of memory bandwidth
    • Hiding memory latency
  • latency hiding
    • Multithreading
    • Prefetching
    • Tradeoffs
  • Interconnection networks
  • Static and dynamic interconnection networks
  • Network interfaces
  • Network topologies: buses
  • Network topologies: crossbar
  • Message passing costs in parrallel computers
• Task decomposition
  • Examples
  • Granularity
  • Degree of concurrency
  • Critical Path Length
  • Limits
  • Task interaction graphs
  • Processes and mapping
  • Decomposition techniques
    • Recursive
    • Data
  • Exploratory decomposition
  • Speculative decomposition
• Symmetric vs Asymmetric Multiprocessing
• Memory Architectures
  • Shared memory
    • UMA
    • NUMA
    • UMA vs NUMA
    • Advantages
    • Disadvantages
  • Distributed memory
    • Advantages
    • Disadvantages
  • Hybrid Distributed Shared Memory
    • Architecture
    • Communication
    • Scalability
    • Draw backs
    • Software Availability
• Parallel Programming Models
  • Shared Memory
  • Threads
  • Message Passing
  • Data Prallel
  • Hybrid
  • Single Program Multiple Data (SPMD)
  • Multiple Program Multiple Data (MPMD)
• OpenMP
  • Complete Chapter# 7 slide