Parallel computing in quantum chemistry /
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Book |
| Language: | English |
| Published: |
Boca Raton :
CRC Press,
[2008]
|
| Subjects: |
Table of Contents:
- I. PARALLEL COMPUTING CONCEPTS AND TERMINOLOGY: 1. Introduction: Parallel computing in Quantum Chemistry: Past and present
- Trends in hardware development
- Moore's law
- Clock speed and performance
- Bandwidth and latency
- Supercomputer performance
- Trends in Parallel software development
- Responding to changes in hardware
- New Algorithms and methods
- New programming models
- 2. Parallel computer architectures: Flynn's classification scheme
- Single-instruction, single-data
- Single-instruction, multiple-data
- Multiple-instruction, multiple-data
- Network architecture
- Direct and indirect networks
- Routing
- Network performance
- Network Topology
- Crossbar
- Ring
- Mesh and Torus
- Hypercube
- Fat tree
- Bus
- Ad Hoc Grid
- Node architecture
- MIMD system architecture
- Memory hierarchy
- Persistent storage
- Local storage
- Network Storage
- Trends in storage
- Reliability
- Homogeneity and heterogeneity
- Commodity versus custom computers
- 3. Communication via Message-Passing: Point-to-point communication operations
- Blocking point-to-point operations
- Non-blocking point-to-point operations
- Collective communication operations
- One-to-all broadcast
- All-to-all broadcast
- All-to-one reduction and all-reduce
- One-sided communication operations
- 4. Multi-threading: Pittfalls of Multi-threading
- Thread-safety
- Comparison of Multi-threading and message-passing
- Hybrid programming
- 5. Parallel performance evaluation: Network performance characteristics
- Performance measures for parallel programs
- Speedup and efficiency
- Scalability
- Performance modeling
- Modeling the execution time
- Performance model example: matrix-vector multiplication
- Presenting and evaluating performance data: a few caveats
- 6. Parallel program design: Distribution of work
- Static task distribution
- Round-Robin and recursive task distributions
- Dynamic task distribution
- Manager-Worker model
- Decentralized task distribution
- Distribution of data
- Designing a communication scheme
- Using collective communication
- Using point-to-point communication
- Design example: Matrix-Vector multiplication
- Using a Row-distributed matrix
- Using a block-distributed matrix
- Summary of key points of parallel program design.
- II. APPLICATIONS OF PARALLEL PROGRAMMING IN QUANTUM CHEMISTRY: 7. Two-electron integral evaluation - - Basics of integral computation
- Parallel implementation using load balancing
- Parallel algorithms distributing shell quartets and pairs
- Performance analysis
- Determination of the load imbalance factor k(p)
- Determination of u and o for integral computation
- Predicted and measured efficiencies
- Parallel implementation using dynamic load balancing
- Parallel algorithm distributing shell pairs
- Performance analysis
- Load imbalance
- Communication time
- Predicted and measured efficiencies
- 8. The Hartree-Fock method: The Hartree-Fock equations
- The Hartree-Fock procedure
- parallel Fock Matrix formation with replicated data
- Parallel Fock matrix formation with distributed data
- 9. Second-Order Moller-Plesset Perturbation theory: The Canonical MP2 equations
- A scalar direct MP2 algorithm
- Parallelization with minimal modifications
- High-performance Parallelization
- Performance of the Parallel algorithms
- 10. Local Moller-Plesset Perturbation theory: The LMP2 equations
- A Scalar LMP2 Algorithm
- Parallel LMP2
- Two-electron integral transformation
- Computation of the residual
- Parallel performance.