diff --git a/src/C-interface/dense/bml_multiply_dense_typed.c b/src/C-interface/dense/bml_multiply_dense_typed.c
index 7d7e8ff0..ac28490b 100644
--- a/src/C-interface/dense/bml_multiply_dense_typed.c
+++ b/src/C-interface/dense/bml_multiply_dense_typed.c
@@ -68,6 +68,7 @@ void TYPED_FUNC(
     MAGMA(gemm) (MagmaNoTrans, MagmaNoTrans,
                  A->N, A->N, A->N, alpha_, B->matrix, B->ld,
                  A->matrix, A->ld, beta_, C->matrix, C->ld, bml_queue());
+    magma_queue_sync(bml_queue());
 #elif defined(MKL_GPU)
     int sizea = A->N * A->N;
     int dnum = 0;
diff --git a/src/C-interface/dense/bml_transpose_dense_typed.c b/src/C-interface/dense/bml_transpose_dense_typed.c
index d3f99de4..42fa6496 100644
--- a/src/C-interface/dense/bml_transpose_dense_typed.c
+++ b/src/C-interface/dense/bml_transpose_dense_typed.c
@@ -50,6 +50,7 @@ bml_matrix_dense_t *TYPED_FUNC(
 #ifdef BML_USE_MAGMA
     MAGMABLAS(transpose) (A->N, A->N, A->matrix, A->ld,
                           B->matrix, B->ld, bml_queue());
+    magma_queue_sync(bml_queue());
 #else
 #ifdef MKL_GPU
 #pragma omp target update from(A_matrix[0:N*N])