Skip to content

Commit 38f87f6

Browse files
committed
Correct assignment of block sizes to loops, don't use fptr table for dispatch.
1 parent 9fe7d6c commit 38f87f6

8 files changed

+604
-46
lines changed

benchmark/openmp.c

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -27,8 +27,8 @@ void cdot3(double* c, double* x, double* A, double* y, long M, long N){
2727
for (long n = 0; n < N; n++){
2828
double tr = 0.0, ti = 0.0;
2929
for(long m = 0; m < M; m++){
30-
tr += x[2*m] * A[2*m + n*N] + x[2*m+1] * A[2*m+1 + n*N];
31-
ti += x[2*m] * A[2*m+1 + n*N] - x[2*m+1] * A[2*m + n*N];
30+
tr += x[2*m] * A[2*m + 2*n*N] + x[2*m+1] * A[2*m+1 + 2*n*N];
31+
ti += x[2*m] * A[2*m+1 + 2*n*N] - x[2*m+1] * A[2*m + 2*n*N];
3232
}
3333
sr += tr * y[2*n ] - ti * y[2*n+1];
3434
si += tr * y[2*n+1] + ti * y[2*n ];
Lines changed: 231 additions & 0 deletions
Loading
81.3 KB
Loading
91.5 KB
Loading
99.3 KB
Loading

0 commit comments

Comments
 (0)