A processor has four floating-point units and runs at 2GHz. Consider the following code:
double a[N], b[N];
for(int i = 0; i < N; i++)
a[i] += b[i];
What is the minimum memory bandwidth in byte/sec that the process must have in order to operate with maximum performance at steady state? N
is a very large number.
My try:
we have in each iteration: 2 loads and 1 store : 3 operations we have 4 FPU : 4x4=16 bytes memory bandwidth = 3x16x2 = 96 GBytes
is my try correct or there is something wrong?
Aucun commentaire:
Enregistrer un commentaire