멀티코어 기말고사 문제 풀이

2023년

1번

• NVCC
• Thrust
• omp_get_thread_num
• block, shared
• guided
• loop variables

2번

1. omp_set_thread_num(NUM)
2. 12
3. private(x)

3번

A

#pragma omp parallel sections

The individual code blocks are distributed over the threads

B

parallel 블록 내부에서만 접근 가능하다.

C

firstprivate variables are initialized to their value before the parallel section

4번

5번

• (i) The serious problem:

  • Too many threads

• (ii) Why?:

  • Excessive parallelism overwhelms resources

2013

1번

1. Synchronization
2. Conditional Variables
3. be blocked / wait
4. independent
5. divide
6. merge
7. data parallel
8. pthread_join
9. pthread_exit

2번

There are several ways in which a pthread may be terminated

• The thread returns from its starting routine (the main routine for the initial thread)
• The thread is canceled is terminated due to making a call to either the exec() or exit()
• The thread makes a call to the pthread_exit subroutine
• If main() finishes first, without calling pthread_exit explicitly itself

5번

#pragma omp parallel for reduction(+:sum) private(x)

2018

1번

1. warp
2. __global__
3. void
4. __host__
5. __device
6. shared memory
7. block
8. private
9. Mutual Exclusion
10. block
11. return
12. pthread_join pthread_exit

2번

• parallel for shared(m, n, a, b, c) private(i, j, sum)

for (i = 0; i<m; i++) {
	int temp = 0;
	for(j = 0; j<n; j++) {
		temp = temp + b[i*N+j] * c[j];
	}
	 a[i] = temp;
}

3번

• blockIdx
• threadIdx
• blockDim
• threadDim

4번

• chunk size n 인 round-robin
• n 개씩 동적으로
• chunk size 가 최소 n까지 줄어든다

5번

왼쪽 q1 찾기
오른쪽 q2 binary search
q3 위치 계산
A[q3] = T[q1]
spawn Par-Merge()
spawn Par-Merge()
sync

6번

• a
• b: cudaMalloc
• c: stencil1D

2017

1번

• General Purpose GPU
• device
• host
• execution configuration (block, thread)
• SIMD
• shared
• loop
• private

2번

• A: thread_id
• B: pthread status code

3번

reduction(+:sum)

4번

• if n ≤ appropriate 숫자 일 때 sorting

5번

• prefix

6번

CUDA!

2016

1번

• block
• shared
• sections
• shared
• loop
• private

2번

3번

4번

• floor (p1 + r1 / 2)

if n >= 30
	sorting other algorithm
else
	

5번

double start = omp_get_wtime()
double end = omp_get_wtime()

2015

1번

• loop
• undefined
• SMID
• execution configuration
  • call by host
  • execute in device
  • return status void
• block
• pthread_join
• pthread_exit

2번

#pragma parallel for reduction(+:sum)

3번

4번

• shared
• destination, source, size, device memory

5번

2014

1번

• shared memory
• warp
• Scalability
• load balancing
• a power of two
• loop variable

2번

3번