We are going to win! wohoo

Dependencies:   mbed mbed-rtos

Revision:
1:6799c07fe510
diff -r 92019d8564a7 -r 6799c07fe510 tvmet/loop/Gemmt.h
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/tvmet/loop/Gemmt.h	Wed Nov 07 14:37:35 2012 +0000
@@ -0,0 +1,115 @@
+/*
+ * Tiny Vector Matrix Library
+ * Dense Vector Matrix Libary of Tiny size using Expression Templates
+ *
+ * Copyright (C) 2001 - 2007 Olaf Petzold <opetzold@users.sourceforge.net>
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ *
+ * $Id: Gemmt.h,v 1.9 2007-06-23 15:58:59 opetzold Exp $
+ */
+
+#ifndef TVMET_LOOP_GEMMT_H
+#define TVMET_LOOP_GEMMT_H
+
+namespace tvmet {
+
+namespace loop {
+
+
+/**
+ * \class gemmt Gemmt.h "tvmet/loop/Gemmt.h"
+ * \brief class for for product matrix-transpose(matrix) operations.
+ *        using formula
+ *        \f[
+ *        M_1\,M_2^{T}
+ *        \f]
+ * \par Example:
+ * \code
+ * template<class T, std::size_t Rows1, std::size_t Cols1, std::size_t Cols2>
+ * inline
+ * void
+ * prod(const Matrix<T, Rows1, Cols1>& lhs, const Matrix<T, Rows2, Cols1>& rhs,
+ *     Matrix<T, Rows1, Rows2>& dest)
+ * {
+ *   for (std::size_t i = 0; i != Rows1; ++i) {
+ *     for (std::size_t j = 0; j != Rows2; ++j) {
+ *       dest(i, j) = tvmet::loop::gemmt<Rows1, Cols1, Cols1>().prod(lhs, rhs, i, j);
+ *     }
+ *   }
+ * }
+ * \endcode
+ * \note The number of cols of rhs matrix have to be equal to cols of rhs matrix.
+ *       The result is a (Rows1 x Rows2) matrix.
+ */
+template<std::size_t Rows1, std::size_t Cols1,
+     std::size_t Cols2 /* unused */>
+class gemmt
+{
+  gemmt(const gemmt&);
+  gemmt& operator=(const gemmt&);
+
+private:
+  enum {
+    count     = Cols1,
+    N         = (count+7)/8
+  };
+
+public:
+  gemmt() { }
+
+public:
+  template<class E1, class E2>
+  static inline
+  typename PromoteTraits<
+    typename E1::value_type,
+    typename E2::value_type
+    >::value_type
+  prod(const E1& lhs, const E2& rhs, std::size_t i, std::size_t j) {
+    typename PromoteTraits<
+      typename E1::value_type,
+      typename E2::value_type
+    >::value_type                  sum(0);
+    std::size_t                 k(0);
+    std::size_t                 n(N);
+
+    // Duff's device
+    switch(count % 8) {
+    case 0: do { sum += lhs(i, k) * rhs(j, k); ++k;
+    case 7:      sum += lhs(i, k) * rhs(j, k); ++k;
+    case 6:      sum += lhs(i, k) * rhs(j, k); ++k;
+    case 5:      sum += lhs(i, k) * rhs(j, k); ++k;
+    case 4:      sum += lhs(i, k) * rhs(j, k); ++k;
+    case 3:      sum += lhs(i, k) * rhs(j, k); ++k;
+    case 2:      sum += lhs(i, k) * rhs(j, k); ++k;
+    case 1:      sum += lhs(i, k) * rhs(j, k); ++k;
+            } while(--n != 0);
+    }
+
+    return sum;
+  }
+};
+
+
+} // namespace loop
+
+} // namespace tvmet
+
+#endif /* TVMET_LOOP_GEMMT_H */
+
+// Local Variables:
+// mode:C++
+// tab-width:8
+// End: