Added prototypes for non-BLAS routines: xSUM and IxMAX (non-absolute counterparts of xASUM and IxAMAX)

2 files changed, 135 insertions, 1 deletions

diff --git a/src/clblast.cc b/src/clblast.cc
index b6efd185..4f2e6fb5 100644
--- a/src/clblast.cc
+++ b/src/clblast.cc
@@ -430,7 +430,32 @@ template StatusCode PUBLIC_API Asum<double2>(const size_t,
                                              const cl_mem, const size_t, const size_t,
                                              cl_command_queue*, cl_event*);
 
-// Index of absolute maxium value in a vector: iSAMAX/iDAMAX/iCAMAX/iZAMAX
+// Sum of values in a vector: SSUM/DSUM/ScSUM/DzSUM
+template <typename T>
+StatusCode Sum(const size_t,
+               cl_mem, const size_t,
+               const cl_mem, const size_t, const size_t,
+               cl_command_queue*, cl_event*) {
+  return StatusCode::kNotImplemented;
+}
+template StatusCode PUBLIC_API Sum<float>(const size_t,
+                                          cl_mem, const size_t,
+                                          const cl_mem, const size_t, const size_t,
+                                          cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Sum<double>(const size_t,
+                                           cl_mem, const size_t,
+                                           const cl_mem, const size_t, const size_t,
+                                           cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Sum<float2>(const size_t,
+                                           cl_mem, const size_t,
+                                           const cl_mem, const size_t, const size_t,
+                                           cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Sum<double2>(const size_t,
+                                            cl_mem, const size_t,
+                                            const cl_mem, const size_t, const size_t,
+                                            cl_command_queue*, cl_event*);
+
+// Index of absolute maximum value in a vector: iSAMAX/iDAMAX/iCAMAX/iZAMAX
 template <typename T>
 StatusCode Amax(const size_t n,
                 cl_mem imax_buffer, const size_t imax_offset,
@@ -461,6 +486,31 @@ template StatusCode PUBLIC_API Amax<double2>(const size_t,
                                              const cl_mem, const size_t, const size_t,
                                              cl_command_queue*, cl_event*);
 
+// Index of maximum value in a vector: iSMAX/iDMAX/iCMAX/iZMAX
+template <typename T>
+StatusCode Max(const size_t,
+               cl_mem, const size_t,
+               const cl_mem, const size_t, const size_t,
+               cl_command_queue*, cl_event*) {
+  return StatusCode::kNotImplemented;
+}
+template StatusCode PUBLIC_API Max<float>(const size_t,
+                                          cl_mem, const size_t,
+                                          const cl_mem, const size_t, const size_t,
+                                          cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Max<double>(const size_t,
+                                           cl_mem, const size_t,
+                                           const cl_mem, const size_t, const size_t,
+                                           cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Max<float2>(const size_t,
+                                           cl_mem, const size_t,
+                                           const cl_mem, const size_t, const size_t,
+                                           cl_command_queue*, cl_event*);
+template StatusCode PUBLIC_API Max<double2>(const size_t,
+                                            cl_mem, const size_t,
+                                            const cl_mem, const size_t, const size_t,
+                                            cl_command_queue*, cl_event*);
+
 // =================================================================================================
 // BLAS level-2 (matrix-vector) routines
 // =================================================================================================
diff --git a/src/clblast_c.cc b/src/clblast_c.cc
index 6e238b77..e6270d57 100644
--- a/src/clblast_c.cc
+++ b/src/clblast_c.cc
@@ -475,6 +475,48 @@ StatusCode CLBlastDzasum(const size_t n,
   return static_cast<StatusCode>(status);
 }
 
+// SUM
+StatusCode CLBlastSsum(const size_t n,
+                       cl_mem asum_buffer, const size_t asum_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Sum<float>(n,
+                                    asum_buffer, asum_offset,
+                                    x_buffer, x_offset, x_inc,
+                                    queue, event);
+  return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastDsum(const size_t n,
+                       cl_mem asum_buffer, const size_t asum_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Sum<double>(n,
+                                     asum_buffer, asum_offset,
+                                     x_buffer, x_offset, x_inc,
+                                     queue, event);
+  return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastScsum(const size_t n,
+                       cl_mem asum_buffer, const size_t asum_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Sum<float2>(n,
+                                     asum_buffer, asum_offset,
+                                     x_buffer, x_offset, x_inc,
+                                     queue, event);
+  return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastDzsum(const size_t n,
+                       cl_mem asum_buffer, const size_t asum_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Sum<double2>(n,
+                                      asum_buffer, asum_offset,
+                                      x_buffer, x_offset, x_inc,
+                                      queue, event);
+  return static_cast<StatusCode>(status);
+}
+
 // AMAX
 StatusCode CLBlastiSamax(const size_t n,
                         cl_mem imax_buffer, const size_t imax_offset,
@@ -517,6 +559,48 @@ StatusCode CLBlastiZamax(const size_t n,
   return static_cast<StatusCode>(status);
 }
 
+// MAX
+StatusCode CLBlastiSmax(const size_t n,
+                       cl_mem imax_buffer, const size_t imax_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Max<float>(n,
+                                    imax_buffer, imax_offset,
+                                    x_buffer, x_offset, x_inc,
+                                    queue, event);
+  return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastiDmax(const size_t n,
+                       cl_mem imax_buffer, const size_t imax_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Max<double>(n,
+                                     imax_buffer, imax_offset,
+                                     x_buffer, x_offset, x_inc,
+                                     queue, event);
+  return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastiCmax(const size_t n,
+                       cl_mem imax_buffer, const size_t imax_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Max<float2>(n,
+                                     imax_buffer, imax_offset,
+                                     x_buffer, x_offset, x_inc,
+                                     queue, event);
+  return static_cast<StatusCode>(status);
+}
+StatusCode CLBlastiZmax(const size_t n,
+                       cl_mem imax_buffer, const size_t imax_offset,
+                       const cl_mem x_buffer, const size_t x_offset, const size_t x_inc,
+                       cl_command_queue* queue, cl_event* event) {
+  auto status = clblast::Max<double2>(n,
+                                      imax_buffer, imax_offset,
+                                      x_buffer, x_offset, x_inc,
+                                      queue, event);
+  return static_cast<StatusCode>(status);
+}
+
 // =================================================================================================
 // BLAS level-2 (matrix-vector) routines
 // =================================================================================================