summaryrefslogtreecommitdiff
path: root/src/routines/level2/xspr2.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/routines/level2/xspr2.cpp')
-rw-r--r--src/routines/level2/xspr2.cpp54
1 files changed, 54 insertions, 0 deletions
diff --git a/src/routines/level2/xspr2.cpp b/src/routines/level2/xspr2.cpp
new file mode 100644
index 00000000..c39a2eb4
--- /dev/null
+++ b/src/routines/level2/xspr2.cpp
@@ -0,0 +1,54 @@
+
+// =================================================================================================
+// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This
+// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max-
+// width of 100 characters per line.
+//
+// Author(s):
+// Cedric Nugteren <www.cedricnugteren.nl>
+//
+// This file implements the Xspr2 class (see the header for information about the class).
+//
+// =================================================================================================
+
+#include "routines/level2/xspr2.hpp"
+
+#include <string>
+
+namespace clblast {
+// =================================================================================================
+
+// Constructor: forwards to base class constructor
+template <typename T>
+Xspr2<T>::Xspr2(Queue &queue, EventPointer event, const std::string &name):
+ Xher2<T>(queue, event, name) {
+}
+
+// =================================================================================================
+
+// The main routine
+template <typename T>
+StatusCode Xspr2<T>::DoSpr2(const Layout layout, const Triangle triangle,
+ const size_t n,
+ const T alpha,
+ const Buffer<T> &x_buffer, const size_t x_offset, const size_t x_inc,
+ const Buffer<T> &y_buffer, const size_t y_offset, const size_t y_inc,
+ const Buffer<T> &ap_buffer, const size_t ap_offset) {
+
+ // Specific Xspr2 functionality is implemented in the kernel using defines
+ return DoHer2(layout, triangle, n, alpha,
+ x_buffer, x_offset, x_inc,
+ y_buffer, y_offset, y_inc,
+ ap_buffer, ap_offset, n,
+ true); // packed matrix
+}
+
+// =================================================================================================
+
+// Compiles the templated class
+template class Xspr2<half>;
+template class Xspr2<float>;
+template class Xspr2<double>;
+
+// =================================================================================================
+} // namespace clblast