31
Example -- shared memory
B := A@east ...;
•In the
shared memory model each processor writes directly into the portion of B that
it ‘owns,’
referencing elements of A as needed
•No
explicit ‘fluff’ regions, but synch needed
barrier_synch(); -- proceed
when all here
for
(i=mylo1_B;i<myhi1_B;i++){
for
(j=mylo2_B;j<myhi2_B;j++){
B[i][j]=A[i][j+1];
}
}
