Question

Given the following code sequence calculating a matrix norm.

double c[96], a[96][96];

for (i=0; i<95; i=++ )

{

c[i] = c[i] + a[i][i]*a[i][i+1];

}

with a and c being arrays of double precision floating point numbers with each element being 8 bytes. No element of a or c is in the cache before executing this code. Please assume row-major ordering. Assuming that the cache is large enough to hold both a and c, and has a cache block size of 64 bytes, determine the number of cache misses (6 pts) and the cache miss rate (4 pts) for this code sequence.

Answer 1

Number of doubles that can be stored in 1 cache block = 64/8 = 8