2.5. High-Performance Computing (HPC)

2.5. High-Performance Computing (HPC)

This area focuses on the principles and practices of developing computational solutions that leverage parallel and distributed computing resources to solve large-scale scientific and engineering problems. It covers computer architectures, parallel programming models, optimization techniques for modern hardware, and the implementation of numerical methods on high-performance systems. This knowledge enables students to tackle computationally intensive problems that are infeasible on single-processor systems.

Table 2.5: List of KUs in the High-Performance Computing area.

2.5.1. HPC/Parallel Computer Architectures  (Core Tier1: 1 hr, Core Tier2: 1 hr) ↑ Back to top

Topics:
Core

  • Flynn's taxonomy and parallel architectures
  • Memory hierarchies and cache coherence
  • Multi-core processors and shared memory
  • Distributed memory systems and clusters
  • Accelerators (GPUs, FPGAs) and heterogeneous computing

Learning Outcomes:
Core:

  1. Classify parallel architectures using Flynn's taxonomy [Familiarity]
  2. Analyze memory hierarchy effects on performance [Assessment]
  3. Compare multi-core and distributed memory systems [Assessment]
  4. Select appropriate accelerators for specific computational tasks [Assessment]

2.5.2. HPC/Shared Memory Parallel Programming  (Core Tier1: 1 hr, Core Tier2: 1 hr) ↑ Back to top

Topics:
Core

  • OpenMP programming model
  • Thread creation and management
  • Synchronization primitives (locks, barriers)
  • Data sharing and scoping
  • Performance optimization and load balancing

Learning Outcomes:
Core:

  1. Implement parallel algorithms using OpenMP [Usage]
  2. Manage threads and task parallelism [Usage]
  3. Apply synchronization mechanisms correctly [Usage]
  4. Optimize shared memory performance [Usage]

2.5.3. HPC/Distributed Memory Parallel Programming  (Core Tier1: 1 hr, Core Tier2: 1 hr) ↑ Back to top

Topics:
Core

  • MPI programming model
  • Point-to-point communication
  • Collective communication operations
  • Process topologies and groups
  • Scalability analysis and optimization

Learning Outcomes:
Core:

  1. Implement distributed algorithms using MPI [Usage]
  2. Handle point-to-point and collective communications [Usage]
  3. Design efficient process topologies [Usage]
  4. Analyze scalability of distributed applications [Assessment]

2.5.4. HPC/Accelerator Programming  (Core Tier1: 1 hr, Core Tier2: 1 hr) ↑ Back to top

Topics:
Core

  • GPU architecture and programming models
  • CUDA programming model
  • OpenACC directives
  • Kernel design and optimization
  • Memory management on accelerators

Learning Outcomes:
Core:

  1. Program GPUs using CUDA [Usage]
  2. Apply OpenACC directives for accelerator programming [Usage]
  3. Design efficient GPU kernels [Usage]
  4. Manage GPU memory hierarchy effectively [Usage]

2.5.5. HPC/Parallel Numerical Methods  (Core Tier1: 1 hr, Core Tier2: 1 hr) ↑ Back to top

Topics:
Core

  • Parallel linear algebra algorithms
  • Parallel PDE solvers
  • Domain decomposition methods
  • Load balancing for numerical computations
  • Performance tuning and profiling

Learning Outcomes:
Core:

  1. Implement parallel numerical algorithms [Usage]
  2. Apply domain decomposition for PDEs [Usage]
  3. Balance computational load across processors [Usage]
  4. Tune performance of parallel numerical codes [Usage]

Spotted a typo, an outdated course, a broken link, or have a suggestion? Let us know.

Scan to open on your phone