IT-803 (D) · Parallel Computing/Unsolved PYQ Paper
IT-803-D Parallel Computing - Jun 2025 Question Paper
Under each question, How to answer lists the points an answer should cover, outlined by CampusPrep from the syllabus. Check the details against your textbook.
-
14 MarksDistinguish among computer terminologies in each of the following groups:
- a)Parallelism versus pipelining
- b)Serial versus parallel processing
- c)Control flow computers versus data flow computers
- d)Uniprocessor systems versus multiprocessors systems
How to answer
- Define parallelism vs pipelining, serial vs parallel, control-flow vs data-flow, uniprocessor vs multiprocessor
- Compare on key aspects: execution model, control, dependency, hardware
- Give one example for each term
-
7 MarksaDescribe atleast four characteristics of MIMD multiprocessors that distinguish them from multiple computer systems or computer networks?
How to answer
- Define MIMD multiprocessors
- List at least four distinguishing characteristics vs multicomputer systems/networks
- Explain tight coupling, shared memory, communication, synchronization
- Give example
-
7 MarksbProve that a k-stage linear pipeline can be at most k times faster than that of a non pipelined serial processor.
How to answer
- Define k-stage linear pipeline timing and non-pipelined timing
- Derive speedup formula S = k*n/(k+n-1)
- Show limit tends to k as n increases
- Conclude at most k times faster
-
7 MarksaWhat are the disadvantages of a cache using Virtual addresses?
How to answer
- Define virtually addressed cache
- List disadvantages: aliasing/synonyms, homonyms, protection, context switch flush
- Explain coherence and translation issues
-
7 MarksbDifferentiate between:
- i)Centralized Vs Distributed Shared caches
- ii)Snooping based vs directory based cache coherence protocol
How to answer
- Define centralized vs distributed shared caches with structure
- Define snooping vs directory based coherence protocols
- Compare on scalability, bandwidth, overhead, suitability
- Give examples
-
7 MarksaDiscuss the architecture of a modern GPU. Give its applications.
How to answer
- Describe modern GPU architecture: SMs, CUDA cores, memory hierarchy
- Explain SIMT execution, interconnect
- List applications: graphics, HPC, AI/ML, scientific computing
-
7 MarksbWhat is the difference between transaction and transactional memory? Explain briefly.
How to answer
- Define transaction vs transactional memory
- Explain hardware vs software transactional memory
- Highlight differences in abstraction, atomicity, implementation
-
7 MarksaDiscuss the architectural characteristics of future systems.
How to answer
- Define future systems trends
- List architectural characteristics: heterogeneity, scalability, energy efficiency
- Explain multicore, accelerators, interconnect
-
7 MarksbBriefly characterize the multicache coherence problem and describe various methods that have been suggested to cope with the problem.
How to answer
- Define multicache coherence problem with example
- Explain write-update, write-invalidate, snooping, directory methods
- Compare methods and coping strategies
-
7 MarksaExplain the four categories of parallelism. What are the steps to design parallel program?
How to answer
- Define four categories of parallelism: job, task, data, pipeline
- Explain each with example
- List steps to design parallel program: partitioning, communication, agglomeration, mapping
-
7 MarksbExplain Distributed-Memory Programming with MPI.
How to answer
- Define distributed-memory model
- Explain MPI concepts: processes, communicators, point-to-point and collective communication
- Describe programming steps with example
-
7 MarksaExplain the performance metric of parallel system.
How to answer
- Define performance metrics: speedup, efficiency, scalability, Amdahl's law
- Give formulas for speedup and efficiency
- Explain factors affecting performance
-
7 MarksbGive a brief note on I/O handling in parallel programming language.
How to answer
- Define I/O challenges in parallel programming
- Explain parallel I/O modes, MPI-IO concepts
- List handling techniques and examples
-
14 MarksWrite brief notes:
- i)Thread Management
- ii)Parallel Data Management
- iii)Fusing Map and scan
How to answer
- Define thread management: creation, synchronization, scheduling
- Explain parallel data management issues
- Explain fusing map and scan operations with example
Go to where you left off?
Quick Add to Notes
Save questions, your own notes and screenshots into notes filed by unit. It takes a free account.
Create free accountHave an account? Log in
Notes Panel