An Autonomoee Institution
ME DEGREE
EXAMINATIONS,APRILMAY 2025
24rCS20l MULTICORE ARCHITECTURE ANDPROGRAMMING
Second Semester
Duration: 3Hour Max Marks 100
Part A
Answer AIIQuestions
10X2-20
1,What is cache coherence in a multicore system?
2. Definc SIMD and MIMD vith examples.
3.What is scalability in parallel computing?
4. Differentiate between mutex and semaphore.
5, List any two OpenMP work-sharing constructs.
6.How is data parallelism handled in OpenMP?
Differentiate bctwccn point-to-point and collcctive
7.
communication
8 Mention any wo MPl performance considerations.
9. Name two case studies for parallel programming applications.
10. List differences between OpenMP and MPI implementations
Part B
Answer All Questions
5X13-65
11. (a) Explainin detail the transition from single-core to multi-core architectureswith examples.
OR
(b)Explain symmetric and distributed shared mernory architectures. Compare both.
12. (a) Explain data race conditions and propose an approach to detect and eliminate them in a complex
multi-threaded environment.
OR
(bYExplain thread communication mechanisms like message queues, pipes, and signals.
13. a) Discuss how functional and loop-levelparallelismare implemented in OpenMP.
OR
(b)Discuss the OpenMP execution model and memory model. Analyze how OpenMP handles thread
creation,data scoping, and synchronization under a shared memory system.
14. (a) Write an MPl program for computing thc sum of an aray using point-to-point communicationand
Cxplain.
OR
(b) Discuss MPI communication mechanismsusing suitable send and receive functions.
15. (a) Compare and contrast tree-based search algorithms using
OpenMP and MPL. Include design
considerations, performance profiling,and communication overhead.
OR
(b)Explain the design and implementation of an n-body solver using OpenMP and MP
Part C
Answer AllQuestions
1X15=15
16. (a) Choosea real-world case study of your own. Design both OpenMP and MPI implementations, then
compare based on scalability, code complexity, and performance.
OR
(b) Design a hybrid parallel solution with MPI and OpenMP for the computational problem matrix
factorization .Provide system architecture, code structure,and performance analysis.