In the experiments thereafter, using $"-03"$ option $($or standard compiler

optimization with compiler other than gcc$)$ when compiling the code. Implement the

original code of matrix multiplication, vary the array size from $N=16$ to $8192$

$($doubling the value of $N$ each time$).$ When does the program execution time

increase sharply? What's the relationship between the cache size of your computer

and the array size around the jump point?