You are on page 1of 5
CONFIDENTIAL, rs CSIJUN 2019/¢8C580 UNIVERSITI TEKNOLOGI MARA FINAL EXAMINATION COURSE : PARALLEL PROCESSING COURSE CODE cSc580 EXAMINATION : JUNE 2019 TIME : 3HOURS INSTRUCTIONS TO CANDIDATES 1 This question paper consists of five (5) questions. 2 ‘Answer ALL questions in the Answer Booklet, Start each answer on a new page. 3 Do not bring any material into the examination room unless permission is given by the invigilator. 4, Please check to make sure that this examination pack consists of |) the Question Paper ji) an Answer Booklet — provided by the Faculty 5. ‘Answer ALL questions in English. DO NOT TURN THIS PAGE UNTIL YOU ARE TOLD TO DO SO This examination paper consists oF 6 printed pages (© Hak Cipta Universit Teknologi MARA CONFIDENTIAL CONFIDENTIAL, 2 CSIJUN 2019/CSC580 QUESTION 1 a) Briefly describe the difference between serial and parallel processing, (4 marks) b) Given the commercial central processing unit (CPU) specifications in Table 1 Table 1 Name Inte!® Pentium® 4 Intel® Core™ 7-97 00KF Processor 681 Processor Base Clock Speed 3.6 GHz 3.6 GHz Number of Core(s) 1 8 Year Released 2006 2019 i) Explain the reason for both CPUs’ base clock speed to remain the same despite the large gap in terms of year released. (2 marks) li) Describe the advantage CPU A has over CPU B. (2 marks) iil) Suggest an enhancement that can be made to allow CPU B to process more data. (2 marks) QUESTION 2 a) Figure 1 shows the execution of 4 instructions in 20 clock cycles. ‘lock ove Figure 4 Each instruction needs to undergo five stages (A, B, C, D, D) to complete the execution i) Illustrate the pipelining execution which will allow the execution above to be completed in earlier clock cycle. (4 marks) ii) Illustrate the superscalar execution that can further improve the throughput in i). (4 marks) ili) State another implicit parallelism technique that can also enhance the execution in i) by taking advantage of instruction level parallelism. lllustrate the execution, (6 marks) (© Hak Cipta Universiti Teknolog! MARA CONFIDENTIAL, CONFIDENTIAL 3 CSIJUN 2019/CSC580 b) Describe TWO (2) advantages of Multiple Instruction Multiple Data Streams (MIMD) over Single Instruction Single Data Streams (SIMD). (4 marks) ©) Explain the difference between static and dynamic interconnection networks. (3 marks) QUESTION 3 a) Define granularity in decomposition. Differentiate between fine-grained and coarse- grained granularity. (6 marks) b) Given the matrix multiplication. Ars Aiz Ass\ (Bir Biz Bas sa Caz Cay Acs Azz Azz ).{ B21 B22 Bos|}=|C21 C22 Coa Isa sz Ass/ \Bs1 Baz Bas. sx C32 Cas. List the tasks that can be decomposed in the above multiplication by using output data decomposition. (9 marks) ©) Figure 2 shows a process mapping for sparse matrix-vector multiplication based on task partitioning, Figure 2 i) List the tasks that are mapped to both processes PO and P1 and tasks that need to be accessed from each process in the mapping above. ( marks) ii) Draw another representation of the above mapping by using sparse matrices. ( marks) il) Show the mapping using the representation in ii) when employing five processes. (6 marks) (© Hak Cipta Universit! Teknologi MARA CONFIDENTIAL, CONFIDENTIAL 4 CSIJUN 2019/CSC580 QUESTION 4 a) Differentiate (with illustration) the differences between non-blocking and blocking message passing operation. (8 marks) b) Describe THREE (3) advantages of threaded programming as compared to message- passing interface. (6 marks) ©) Table 2 shows the execution time of a program using different number of processes in a processor with six processing cores. Table 2 co Semenneeeen ‘Sequential 6000 2 3400 2500 2000 1800 1700 1720 1735 0,1 olen} i) Calculate the speedup and efficiency for every execution. (4 marks) ii) Discuss the reason behind the decline in terms of speedup when the program is ‘executed using more than six processes. (2 marks) © Hak Cipta Universiti Teknologi MARA CONFIDENTIAL, CONFIDENTIAL, 5 CSIJUN 2019/CSC580 QUESTION 5 Consider the undirected weighted graph in Figure 3. Figure 3 a) Find the minimum spanning tree of the graph using Kruskal's algorithm. (12 marks) b) Show the parallel formulation for the above algorithm. (8 marks) END OF QUESTION PAPER (© Hak Cipta Universit! Teknolog! MARA CONFIDENTIAL,

You might also like