Quiznetik

High Performance Computing (HPC) | Set 1

1. A CUDA program is comprised of two primary components: a host and a _____.

Correct : A. gpu??kernel

2. The kernel code is dentified by the ________qualifier with void return type

Correct : B. __global__??

3. The kernel code is only callable by the host

Correct : A. true

4. The kernel code is executable on the device and host

Correct : B. false

5. Calling a kernel is typically referred to as _________.

Correct : D. kernel invocation

6. Host codes in a CUDA application can Initialize a device

Correct : A. true

7. Host codes in a CUDA application can Allocate GPU memory

Correct : A. true

8. Host codes in a CUDA application can not Invoke kernels

Correct : B. false

9. CUDA offers the Chevron Syntax to configure and execute a kernel.

Correct : A. true

10. the BlockPerGrid and ThreadPerBlock parameters are related to the ________ model supported by CUDA.

Correct : C. thread??abstraction

11. _________ is Callable from the device only

Correct : C. _device_

12. ______ is Callable from the host

Correct : B. __global__??

13. ______ is Callable from the host

Correct : A. _host_

14. CUDA supports ____________ in which code in a single thread is executed by all other threads.

Correct : C. thread abstraction

15. In CUDA, a single invoked kernel is referred to as a _____.

Correct : C. grid

16. A grid is comprised of ________ of threads.

Correct : A. block

17. A block is comprised of multiple _______.

Correct : A. treads

18. a solution of the problem in representing the parallelismin algorithm is

Correct : D. cuda

19. Host codes in a CUDA application can not Reset a device

Correct : B. false

20. Host codes in a CUDA application can Transfer data to and from the device

Correct : A. true

21. Host codes in a CUDA application can not Deallocate memory on the GPU

Correct : B. false

22. Any condition that causes a processor to stall is called as _____.

Correct : A. hazard

23. The time lost due to branch instruction is often referred to as _____.

Correct : C. branch penalty

24. _____ method is used in centralized systems to perform out of order execution.

Correct : B. score boarding

25. The computer cluster architecture emerged as an alternative for ____.

Correct : C. super computers

26. NVIDIA CUDA Warp is made up of how many threads?

Correct : D. 32

27. Out-of-order instructions is not possible on GPUs.

Correct : B. false

28. CUDA supports programming in ....

Correct : C. c, c++, third party wrappers for java, python, and more

29. FADD, FMAD, FMIN, FMAX are ----- supported by Scalar Processors of NVIDIA GPU.

Correct : A. 32-bit ieee floating point instructions

30. Each streaming multiprocessor (SM) of CUDA herdware has ------ scalar processors (SP).

Correct : D. 8

31. Each NVIDIA GPU has ------ Streaming Multiprocessors

Correct : D. 16

32. CUDA provides ------- warp and thread scheduling. Also, the overhead of thread creation is on the order of ----.

Correct : B. “zero-overhead”, 1 clock

33. Each warp of GPU receives a single instruction and “broadcasts” it to all of its threads. It is a ---- operation.

Correct : B. simt (single instruction multiple thread)

34. Limitations of CUDA Kernel

Correct : B. no recursion, no call stack, no static variable declarations

35. What is Unified Virtual Machine

Correct : A. it is a technique that allow both cpu and gpu to read from single virtual machine, simultaneously.

36. _______ became the first language specifically designed by a GPU Company to facilitate general purpose computing on ____.

Correct : C. cuda c, gpus.

37. The CUDA architecture consists of --------- for parallel computing kernels and functions.

Correct : D. ptx instruction set architecture

38. CUDA stands for --------, designed by NVIDIA.

Correct : C. compute unified device architecture

39. The host processor spawns multithread tasks (or kernels as they are known in CUDA) onto the GPU device. State true or false.

Correct : A. true

40. The NVIDIA G80 is a ---- CUDA core device, the NVIDIA G200 is a ---- CUDA core device, and the NVIDIA Fermi is a ---- CUDA core device.

Correct : A. 128, 256, 512

41. NVIDIA 8-series GPUs offer -------- .

Correct : A. 50-200 gflops

42. IADD, IMUL24, IMAD24, IMIN, IMAX are ----------- supported by Scalar Processors of NVIDIA GPU.

Correct : B. 32-bit integer instructions

43. CUDA Hardware programming model supports: a) fully generally data-parallel archtecture; b) General thread launch; c) Global load-store; d) Parallel data cache; e) Scalar architecture; f) Integers, bit operation

Correct : D. a,b,c,d,e,f

44. In CUDA memory model there are following memory types available: a) Registers; b) Local Memory; c) Shared Memory; d) Global Memory; e) Constant Memory; f) Texture Memory.

Correct : C. a, b, c, d, e, f

45. What is the equivalent of general C program with CUDA C: int main(void) { printf("Hello, World!\n"); return 0; }

Correct : B. __global__ void kernel( void ) { } int main ( void ) { kernel <<<1,1>>>(); printf("hello, world!\\n"); return 0; }

46. Which function runs on Device (i.e. GPU): a) __global__ void kernel (void ) { } b) int main ( void ) { ... return 0; }

Correct : A. a

47. A simple kernel for adding two integers: __global__ void add( int *a, int *b, int *c ) { *c = *a + *b; } where __global__ is a CUDA C keyword which indicates that:

Correct : A. add() will execute on device, add() will be called from host

48. If variable a is host variable and dev_a is a device (GPU) variable, to allocate memory to dev_a select correct statement:

Correct : C. cudamalloc( (void**) &dev_a, sizeof( int ) )

49. If variable a is host variable and dev_a is a device (GPU) variable, to copy input from variable a to variable dev_a select correct statement:

Correct : B. cudamemcpy( dev_a, &a, size, cudamemcpyhosttodevice );

50. Triple angle brackets mark in a statement inside main function, what does it indicates?

Correct : A. a call from host code to device code

51. What makes a CUDA code runs in parallel

Correct : D. first parameter value inside triple angle bracket (n) indicates excecution of kernel n times in parallel

52. In ___________, the number of elements to be sorted is small enough to fit into the process's main memory.

Correct : A. internal sorting

53. ______________ algorithms use auxiliary storage (such as tapes and hard disks) for sorting because the number of elements to be sorted is too large to fit into memory.

Correct : C. external sorting

54. ______ can be comparison-based or noncomparison-based.

Correct : B. sorting

55. The fundamental operation of comparison-based sorting is ________.

Correct : A. compare-exchange

56. The complexity of bubble sort is Θ(n2).

Correct : A. true

57. Bubble sort is difficult to parallelize since the algorithm has no concurrency.

Correct : A. true

58. Quicksort is one of the most common sorting algorithms for sequential computers because of its simplicity, low overhead, and optimal average complexity.

Correct : A. true

59. The performance of quicksort depends critically on the quality of the ______-.

Correct : B. pivot

60. the complexity of quicksort is O(nlog n).

Correct : A. true

61. The main advantage of ______ is that its storage requirement is linear in the depth of the state space being searched.

Correct : B. dfs

62. _____ algorithms use a heuristic to guide search.

Correct : A. bfs

63. If the heuristic is admissible, the BFS finds the optimal solution.

Correct : A. true

64. The search overhead factor of the parallel system is defined as the ratio of the work done by the parallel formulation to that done by the sequential formulation

Correct : A. true

65. The critical issue in parallel depth-first search algorithms is the distribution of the search space among the processors.

Correct : A. true

66. Graph search involves a closed list, where the major operation is a _______

Correct : C. lookup

67. Breadth First Search is equivalent to which of the traversal in the Binary Trees?

Correct : C. level-order traversal

68. Time Complexity of Breadth First Search is? (V – number of vertices, E – number of edges)

Correct : A. o(v + e)

69. Which of the following is not an application of Breadth First Search?

Correct : B. when the graph is a linked list

70. In BFS, how many times a node is visited?

Correct : C. equivalent to number of indegree of the node

71. Is Best First Search a searching algorithm used in graphs.

Correct : A. true

72. Which of the following is not a stable sorting algorithm in its typical implementation.

Correct : C. quick sort

73. Which of the following is not true about comparison based sorting algorithms?

Correct : D. heap sort is not a comparison based sorting algorithm.

74. mathematically efficiency is

Correct : A. e=s/p

75. Cost of a parallel system is sometimes referred to____ of product

Correct : C. both

76. Scaling Characteristics of Parallel Programs Ts is

Correct : B. constant

77. Speedup tends to saturate and efficiency _____ as a consequence of Amdahl’s law.

Correct : C. decreases

78. Speedup obtained when the problem size is _______ linearly with the number of processing elements.

Correct : A. increase

79. The n × n matrix is partitioned among n processors, with each processor storing complete ___ of the matrix.

Correct : A. row

80. cost-optimal parallel systems have an efficiency of ___

Correct : A. 1

81. The n × n matrix is partitioned among n2 processors such that each processor owns a _____ element.

Correct : C. single

82. how many basic communication operations are used in matrix vector multiplication

Correct : C. 3

83. In DNS algorithm of matrix multiplication it used

Correct : C. 3d partition

84. In the Pipelined Execution, steps contain

Correct : D. all

85. the cost of the parallel algorithm is higher than the sequential run time by a factor of __

Correct : A. 2020-03-02 00:00:00

86. The load imbalance problem in Parallel Gaussian Elimination: can be alleviated by using a ____ mapping

Correct : B. cyclic

87. A parallel algorithm is evaluated by its runtime in function of

Correct : D. all

88. For a problem consisting of W units of work, p__W processors can be used optimally.

Correct : A. <=

89. C(W)__Θ(W) for optimality (necessary condition).

Correct : D. equals

90. many interactions in oractical parallel programs occur in _____ pattern

Correct : A. well defined

91. efficient implementation of basic communication operation can improve

Correct : A. performance

92. efficient use of basic communication operations can reduce

Correct : A. development effort and

93. Group communication operations are built using_____ Messenging primitives.

Correct : A. point-to-point

94. one processor has a piece of data and it need to send to everyone is

Correct : A. one -to-all

95. the dual of one -to-all is

Correct : A. all-to-one reduction

96. Data items must be combined piece-wise and the result made available at

Correct : A. target processor finally

97. wimpleat way to send p-1 messages from source to the other p-1 processors

Correct : C. concurrency

98. In a eight node ring, node ____ is source of broadcast

Correct : D. 0

99. The processors compute ______ product of the vector element and the loval matrix

Correct : A. local

100. one to all broadcast use

Correct : A. recursive doubling