c++ - Cache Poisoning Issue for deep nested loop -

i writing code mathematical method (incomplete cholesky) , have hit curious roadblock. please see following simplified code.

for(k=0;k<nosunknowns;k++) {     //pieces of code     for(i=k+1;i<nosunknowns;i++)     {         // more code     }     for(j=k+1;j<nosunknowns;j++)     {         for(i=j;i<nosunknowns;i++)         {             //some more code              if(xok && yok && zok)             {                 if(xdf == 1 && ydf == 0 && zdf == 0)                 {                     for(row=0;row<3;row++)                     {                         for(col=0;col<3;col++)                         {                             // 3x3 static arrays line                             statobj->a1_[row][col] -= localfuncarr[row][col];                         }                     }                 }             }         }//inner loop ends here     }//inner loop j ends here }//outer loop k ends here 

for context,

statobj object containing number of 3x3 static double arrays. initializing statobj call new function. populating arrays inside using mathematical functions. 1 such array a1_. value of variable nosunknowns around 3000. array localfuncarr generated matrix multiplication , double array.

now problem:

1. when use line shown in code, code runs extremely sluggishly. 245secs whole function.

2. when comment out said line, code performs extremely fast. takes 6 secs.

3. now when replace said line following line : localfuncarr[row][col] += 3.0, again code runs same speed of case(2) above.

clearly call statobj->a1_ making code run slow.

my question(s):

1. is cache poisoning reason why happening ?

2. if so, changed in terms of array initialization/object initialization/loop unrolling or matter form of code optimization can speed ?

any insights experienced folks highly appreciated.

edit: changed description more verbose , redress of points mentioned in comments.

if conditions true, line of code executed 3000x3000x3000x3x3 times. that's 245 billion times. depending on hardware architecture 245 seconds might reasonable timing (that's 1 iteration every 2 cycles - assuming 2ghz processor). in case there isn't in code suggests cache poisoning.