Mumbai University · Final Year Computer Engineering · Rev-2019 'C' Scheme
High Performance Computing
CSDC8022 · Department Optional Course 6 · Semester VIII · 3 credits · Internal tests 20 (average of two) + end-semester 80
End-semester paper (3 hours): Six questions of 20 marks; you answer four. Each module's share of the paper follows its lecture hours.
One of the Department Optional Course 6 options.
Module 1: Introduction to Parallel Computing
5 hoursLevels of parallelism; Flynn's, Shore's, Feng's and Händler's classifications; distributed, shared and hybrid memory; arithmetic pipelines and array processors.
Examined in the 80-mark end-semester paper (about 14% of it, by lecture hours); the class tests cover what has been taught by then
- Levels of parallelism
- Classifications of parallel architectures
- Shared, distributed and hybrid memory
Module 2: Platforms and Algorithm Design
11 hoursPhysical organisation of parallel platforms; communication costs; decomposition; tasks and interactions; mapping for load balancing; containing interaction overheads; parallel algorithm models.
Examined in the 80-mark end-semester paper (about 30% of it, by lecture hours); the class tests cover what has been taught by then
- Communication costs in parallel machines
- Decomposition techniques
- Mapping for load balancing
- Parallel algorithm models
Module 3: Performance Measures
3 hoursSpeedup, efficiency, cost and scalability; granularity; Amdahl's and Gustafson's laws; bottlenecks; the Karp–Flatt metric.
Examined in the 80-mark end-semester paper (about 8% of it, by lecture hours); the class tests cover what has been taught by then
- Speedup, efficiency and scalability
- Amdahl's and Gustafson's laws
Module 4: OpenMP and MPI
10 hoursThreads and the fork–join model; parallel, for and schedule directives; private and shared variables; critical, atomic, barrier and reduction; MPI send, receive and barrier; hybrid MPI with OpenMP.
Examined in the 80-mark end-semester paper (about 27% of it, by lecture hours); the class tests cover what has been taught by then
- OpenMP directives
- Synchronisation in OpenMP
- MPI point-to-point communication
- Hybrid MPI and OpenMP programming
Module 5: Accelerators
4 hoursGPGPUs; introduction to CUDA; heterogeneous computing with OpenCL.
Examined in the 80-mark end-semester paper (about 11% of it, by lecture hours); the class tests cover what has been taught by then
- GPGPU and CUDA
- OpenCL
Module 6: HPC in the Cloud
4 hoursVirtualisation and containers; parallel computing frameworks; scaling; cloud HPC use cases.
Examined in the 80-mark end-semester paper (about 11% of it, by lecture hours); the class tests cover what has been taught by then
- HPC in the cloud
Module contents are summarised from the official Mumbai University syllabus . AiCanCode.org is not affiliated with the University of Mumbai.