Home
About
Services
Research
News
Events
Contact

   
Virtual Workshop Modules
Distributed Memory Programming
Introduction
2. Trends
    2.1.1 Technologies
    2.1.2 Internal Parallel
    2.1.3 Clock Rates
    2.1.4 Memory Rates
    2.1.5 Multilevel
    2.1.6 Interleaved
    2.2 Shared Memory
    2.3 Distributed Memory
3. On the Horizon
4. Time Hierarchy
5. Memory Architectures
    5.1 Multicomputers
    5.2 Distributed Shared
6. Programming Models
    6.1 Data Parallel
    6.2 Message Passing
    6.3 Active Messages
7. Strengths, Weaknesses
8. Key Principles
    8.1 Program Hygiene
    8.2 Addressability
    8.3 Communication
    8.4 Collective Operations
    8.5 Blocking
    8.6 Non-Blocking
    8.7 Synchronization
    8.8 Pitfalls
9. Considerations
    9.1 Owner Rule
    9.2 Strategies
    9.3 Scaling
    9.4 Computation
    9.5 Load Balance
    9.6 Models
    9.7 Models (continued)
    9.8 Conclusions
    9.9 Speedup
10. Conclusion
References
Quiz
Evaluation
Questions?
2.1.4 Memory Rates

Rates

Memory technology has, unfortunately, not experienced the same marked degree of improvement, at least in access time, as processor technology has in execution capability. Memory access has been, and will continue to be for the foreseeble future, the single largest bottleneck to significant advances in computational throughput.


© 2008 Cornell University | Cornell University Center for Advanced Computing | Copyright Statement