RTMNU · Third Year B.Tech Computer Science & Engineering (AI & ML) · NEP 2020
Parallel Computer Architecture and Programming
BCSEAI&ML5T18.1 · Elective I · Semester V · 3 credits · College 30 + university 70
One of the Elective I options.
Unit 1: Introduction & Technique of Parallelism
7 hoursTrends towards parallel computing and parallelism in uniprocessors; architectural classification schemes; Amdahl's and Moore's laws and scalable performance; parallel processing in memory, parallel algorithms and their complexity; models of parallel processing; cache coherence and its protocols.
Examined in the 70-mark end-semester paper
- Parallel and distributed computing
- Flynn's taxonomy
- Amdahl's and Gustafson's laws
- Performance metrics and scaling
- Parallel algorithm models
- Shared memory and cache coherence
Unit 2: Pipeline & Vector Processing
7 hoursData and resource dependences; control-flow versus data-flow computers; linear pipelining, classification of pipeline processors, general pipelines and reservation tables; instruction and arithmetic pipelines; vector processing and the Cray-1; pipeline hazards; VLIW computers and array processing.
Examined in the 70-mark end-semester paper
- Conditions of parallelism: data and resource dependences
- Control-flow versus data-flow computers
- Instruction pipelining: throughput and speedup
- Reservation tables and arithmetic pipelines
- Pipeline hazards and their solutions
- Vector processing, VLIW and array processors
Unit 3: Parallel Models
7 hoursPRAM and basic algorithms, data broadcasting, parallel prefix computation, shared-memory algorithms and parallel selection; circuit-level examples, the tree-structured dictionary machine and parallel prefix networks; sorting and routing on a 2D mesh or torus, data-routing operations and greedy routing.
Examined in the 70-mark end-semester paper
- The PRAM model and shared-memory algorithms
- Parallel prefix sum
- Sorting and routing on a 2D mesh
- Data routing operations and greedy routing
Unit 4: Mesh-Based Architectures
8 hours2D mesh algorithms, meshes of trees and low-diameter architectures; hypercubes and their algorithms, sorting and routing on hypercubes, bitonic sorting, dimension-order routing and broadcasting; butterfly and permutation networks, plus-or-minus-2^i networks, cube-connected cycles and shuffle-exchange networks.
Examined in the 70-mark end-semester paper
- Meshes, meshes of trees and low-diameter architectures
- Hypercube algorithms: sorting, routing and broadcasting
- Bitonic sorting networks
- Butterfly, cube-connected cycles and shuffle-exchange networks
Unit 5: Multiprocessor Architecture and Programming
7 hoursEmulations among architectures and scheduling; distributed shared memory; data storage, input and output; multithreading and latency hiding; parallel I/O; defect-, fault- and error-level methods; parallel programming, parallel operating systems and parallel file systems.
Examined in the 70-mark end-semester paper
- Emulation and scheduling among parallel architectures
- Distributed shared memory
- Multithreading and latency hiding
- Parallel I/O and fault tolerance
- Parallel operating systems and file systems
Unit contents are summarised from the official RTMNU syllabus . AiCanCode.org is not affiliated with Rashtrasant Tukadoji Maharaj Nagpur University (RTMNU).