Make main CTE loop over columns dynamic rather than static

This should reduce possible runtimes being held up by contended cores when run as the default 'greedy' mode. resolves #152 Signed-off-by: James Noss <[email protected]>
spacetelescope · Jun 15, 2017 · afae94f · afae94f
1 parent fa6b200
commit afae94f
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/ctegen2/ctegen2.c b/ctegen2/ctegen2.c
@@ -97,7 +97,7 @@ int inverseCTEBlur(const SingleGroup * input, SingleGroup * output, SingleGroup
             Bool localOK = True;
             {unsigned j;
 #ifdef _OPENMP
-            #pragma omp for schedule(static)
+            #pragma omp for schedule(dynamic)
 #endif
             for (j = 0; j < nColumns; ++j)
             {