Merge pull request #1895 from eeprude/axpby_improvement

Axpby: improvement on unification attempt logic and on the execution of a diversity of situations
kokkos · Oct 23, 2023 · 10cbf87 · 10cbf87
2 parents f8aa101 + 12d1fd4
commit 10cbf87
Show file tree

Hide file tree

Showing 11 changed files with 4,748 additions and 1,188 deletions.
diff --git a/blas/impl/KokkosBlas1_axpby_impl.hpp b/blas/impl/KokkosBlas1_axpby_impl.hpp
diff --git a/blas/impl/KokkosBlas1_axpby_mv_impl.hpp b/blas/impl/KokkosBlas1_axpby_mv_impl.hpp
diff --git a/blas/impl/KokkosBlas1_axpby_spec.hpp b/blas/impl/KokkosBlas1_axpby_spec.hpp
diff --git a/blas/impl/KokkosBlas1_axpby_unification_attempt_traits.hpp b/blas/impl/KokkosBlas1_axpby_unification_attempt_traits.hpp
diff --git a/blas/src/KokkosBlas1_axpby.hpp b/blas/src/KokkosBlas1_axpby.hpp
diff --git a/blas/unit_test/Test_Blas.hpp b/blas/unit_test/Test_Blas.hpp
@@ -24,6 +24,7 @@
 #include "Test_Blas1_asum.hpp"
 #include "Test_Blas1_axpby.hpp"
 #include "Test_Blas1_axpy.hpp"
+#include "Test_Blas1_axpby_unification.hpp"
 #include "Test_Blas1_dot.hpp"
 #include "Test_Blas1_iamax.hpp"
 #include "Test_Blas1_mult.hpp"

diff --git a/blas/unit_test/Test_Blas1_axpby.hpp b/blas/unit_test/Test_Blas1_axpby.hpp
@@ -109,8 +109,6 @@ void impl_test_axpby_mv(int N, int K) {
   Kokkos::deep_copy(org_y.h_base, y.d_base);
   Kokkos::deep_copy(x.h_base, x.d_base);
 
-  Kokkos::View<ScalarB*, Kokkos::HostSpace> r("Dot::Result", K);
-
   KokkosBlas::axpby(a, x.d_view, b, y.d_view);
   Kokkos::deep_copy(y.h_base, y.d_base);