From b2bc5d6becd479b84561781bc5a0beea40c476b7 Mon Sep 17 00:00:00 2001
From: Luc Berger-Vergiat <lberge@sandia.gov>
Date: Mon, 15 Nov 2021 06:54:43 -0700
Subject: [PATCH 1/2] Fix issues with SpMV using tensor cores

This might be a bit aggressive but it seems
to pass build and tests correctly. Basically
removing compilation of the tensore core
code path unless it is compiled with the cuda
compiler.
---
 src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp | 2 +-
 src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp | 8 +++++---
 2 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp
index 1e5c098814..cbec458ae9 100644
--- a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp
+++ b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp
@@ -45,7 +45,7 @@
 #ifndef KOKKOSSPARSE_IMPL_SPMV_TENSOR_CORE_DEF_HPP_
 #define KOKKOSSPARSE_IMPL_SPMV_TENSOR_CORE_DEF_HPP_
 
-#if defined(KOKKOS_ENABLE_CUDA) && \
+#if defined(KOKKOS_ENABLE_CUDA) && defined(__CUDA_ARCH__) &&	\
     (defined(KOKKOS_ARCH_VOLTA) || defined(KOKKOS_ARCH_AMPERE))
 
 #include <type_traits>
diff --git a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp
index 621c52c157..530f713044 100644
--- a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp
+++ b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp
@@ -50,7 +50,7 @@
 
 #include "KokkosSparse_BsrMatrix.hpp"
 #include "KokkosKernels_Controls.hpp"
-#if !defined(KOKKOSKERNELS_ETI_ONLY) || KOKKOSKERNELS_IMPL_COMPILE_LIBRARY
+#if !defined(KOKKOSKERNELS_ETI_ONLY) || KOKKOSKERNELS_IMPL_COMPILE_LIBRARY && defined(__CUDA_ARCH__)
 #include <KokkosSparse_spmv_bsrmatrix_impl.hpp>
 #endif
 
@@ -136,7 +136,9 @@ struct SPMV_BSRMATRIX<AT, AO, AD, AM, AS, XT, XL, XD, XM, YT, YL, YD, YM,
       }
     }
 
-#if defined(KOKKOS_ARCH_AMPERE)
+    // If not compiling with the cuda compiler, no need to try
+    // to compile the wmma functions...
+#if defined(KOKKOS_ARCH_AMPERE) && defined(__CUDA_ARCH__)
     typedef typename XVector::non_const_value_type XScalar;
     typedef typename AMatrix::non_const_value_type AScalar;
     typedef Kokkos::Experimental::half_t Half;
@@ -173,7 +175,7 @@ struct SPMV_BSRMATRIX<AT, AO, AD, AM, AS, XT, XL, XD, XM, YT, YL, YD, YM,
                                         4>::dispatch(alpha, A, x, beta, y);
     }
 
-#elif defined(KOKKOS_ARCH_VOLTA)
+#elif defined(KOKKOS_ARCH_VOLTA) && defined(__CUDA_ARCH__)
     /* Volta has float += half * half
        use it for all matrices
     */

From 2f71c8c37fec2fccfa2938ef09dc3ce4995635e4 Mon Sep 17 00:00:00 2001
From: Luc Berger-Vergiat <lberge@sandia.gov>
Date: Mon, 15 Nov 2021 06:58:43 -0700
Subject: [PATCH 2/2] Apply clang-format

---
 src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp | 2 +-
 src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp
index cbec458ae9..f4254f129d 100644
--- a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp
+++ b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_impl.hpp
@@ -45,7 +45,7 @@
 #ifndef KOKKOSSPARSE_IMPL_SPMV_TENSOR_CORE_DEF_HPP_
 #define KOKKOSSPARSE_IMPL_SPMV_TENSOR_CORE_DEF_HPP_
 
-#if defined(KOKKOS_ENABLE_CUDA) && defined(__CUDA_ARCH__) &&	\
+#if defined(KOKKOS_ENABLE_CUDA) && defined(__CUDA_ARCH__) && \
     (defined(KOKKOS_ARCH_VOLTA) || defined(KOKKOS_ARCH_AMPERE))
 
 #include <type_traits>
diff --git a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp
index 530f713044..10abd991d1 100644
--- a/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp
+++ b/src/sparse/impl/KokkosSparse_spmv_bsrmatrix_spec.hpp
@@ -50,7 +50,8 @@
 
 #include "KokkosSparse_BsrMatrix.hpp"
 #include "KokkosKernels_Controls.hpp"
-#if !defined(KOKKOSKERNELS_ETI_ONLY) || KOKKOSKERNELS_IMPL_COMPILE_LIBRARY && defined(__CUDA_ARCH__)
+#if !defined(KOKKOSKERNELS_ETI_ONLY) || \
+    KOKKOSKERNELS_IMPL_COMPILE_LIBRARY && defined(__CUDA_ARCH__)
 #include <KokkosSparse_spmv_bsrmatrix_impl.hpp>
 #endif