Merge tag '1.6.0' into gspr/post-bookworm

4 files changed, 36 insertions, 17 deletions

diff --git a/scripts/benchmark/plot.py b/scripts/benchmark/plot.py
index b0b63df3..2cf5d04d 100644
--- a/scripts/benchmark/plot.py
+++ b/scripts/benchmark/plot.py
@@ -76,7 +76,12 @@ def plot_graphs(results, file_name, num_rows, num_cols,
         for col in range(num_cols):
             index = row * num_cols + col
             result = results[index]
-            ax = axes[row, col]
+            if num_rows == 1:
+                ax = axes[col]
+            elif num_cols == 1:
+                ax = axes[row]
+            else:
+                ax = axes[row, col]
             plt.sca(ax)
             print("[plot] Plotting subplot %d" % index)
 
@@ -90,7 +95,8 @@ def plot_graphs(results, file_name, num_rows, num_cols,
                 x_ticks = [v if not (i % 2) else "" for i, v in enumerate(x_ticks)]
 
             # Sets the y-data
-            y_list = [[r[y_key] if y_key in r.keys() else 0 for r in result] for y_key in y_keys[index]]
+            y_list = [[r[y_key] if y_key in r.keys() and not isinstance(r[y_key], str) else 0 for r in result]
+                      for y_key in y_keys[index]]
             y_max = [max(y) if len(y) else 1 for y in y_list]
             y_max = max(y_max) if len(y_list) > 0 else 1
 
diff --git a/scripts/generator/generator.py b/scripts/generator/generator.py
index 875b1a55..3d66960f 100755
--- a/scripts/generator/generator.py
+++ b/scripts/generator/generator.py
@@ -59,13 +59,13 @@ ald_m = "The value of `a_ld` must be at least `m`."
 ald_n = "The value of `a_ld` must be at least `n`."
 ald_k_one = "The value of `a_ld` must be at least `k + 1`."
 ald_kl_ku_one = "The value of `a_ld` must be at least `kl + ku + 1`."
-ald_transa_m_k = "When `transpose_a == Transpose::kNo`, then `a_ld` must be at least `m`, otherwise `a_ld` must be at least `k`."
-ald_trans_n_k = "When `transpose == Transpose::kNo`, then `a_ld` must be at least `n`, otherwise `a_ld` must be at least `k`."
+ald_transa_m_k = "When `(transpose_a == Transpose::kNo && layout == Layout::kColMajor) || (transpose_a == Transpose::kYes && layout == Layout::kRowMajor)`, then `a_ld` must be at least `m`, otherwise `a_ld` must be at least `k`."
+ald_trans_n_k = "When `(transpose == Transpose::kNo && layout == Layout::kColMajor) || (transpose == Transpose::kYes && layout == Layout::kRowMajor)`, then `a_ld` must be at least `n`, otherwise `a_ld` must be at least `k`."
 ald_side_m_n = "When `side = Side::kLeft` then `a_ld` must be at least `m`, otherwise `a_ld` must be at least `n`."
 bld_m = "The value of `b_ld` must be at least `m`."
 bld_n = "The value of `b_ld` must be at least `n`."
-bld_transb_k_n = "When `transpose_b == Transpose::kNo`, then `b_ld` must be at least `k`, otherwise `b_ld` must be at least `n`."
-bld_trans_n_k = "When `transpose == Transpose::kNo`, then `b_ld` must be at least `n`, otherwise `b_ld` must be at least `k`."
+bld_transb_k_n = "When `(transpose_b == Transpose::kNo && layout == Layout::kColMajor) || (transpose_b == Transpose::kYes && layout == Layout::kRowMajor)`, then `b_ld` must be at least `k`, otherwise `b_ld` must be at least `n`."
+bld_trans_n_k = "When `(transpose == Transpose::kNo && layout == Layout::kColMajor) || (transpose == Transpose::kYes && layout == Layout::kRowMajor)`, then `b_ld` must be at least `n`, otherwise `b_ld` must be at least `k`."
 cld_m = "The value of `c_ld` must be at least `m`."
 cld_n = "The value of `c_ld` must be at least `n`."
 
diff --git a/scripts/generator/generator/cpp.py b/scripts/generator/generator/cpp.py
index 6dc3fc93..e32738ee 100644
--- a/scripts/generator/generator/cpp.py
+++ b/scripts/generator/generator/cpp.py
@@ -226,7 +226,10 @@ def wrapper_clblas(routine):
 
                 # Convert to float (note: also integer buffers are stored as half/float)
                 for buf in routine.inputs + routine.outputs:
-                    result += "  auto " + buf + "_buffer_bis = HalfToFloatBuffer(" + buf + "_buffer, queues[0]);" + NL
+                    if buf not in routine.index_buffers():
+                        result += "  auto " + buf + "_buffer_bis = HalfToFloatBuffer(" + buf + "_buffer, queues[0]);" + NL
+                    else:
+                        result += "  auto " + buf + "_buffer_bis = " + buf + "_buffer;" + NL
 
                 # Call the float routine
                 result += "  auto status = clblasX" + routine.name + "("
@@ -236,7 +239,8 @@ def wrapper_clblas(routine):
 
                 # Convert back to half
                 for buf in routine.outputs:
-                    result += "  FloatToHalfBuffer(" + buf + "_buffer, " + buf + "_buffer_bis, queues[0]);" + NL
+                    if buf not in routine.index_buffers():
+                        result += "  FloatToHalfBuffer(" + buf + "_buffer, " + buf + "_buffer_bis, queues[0]);" + NL
                 result += "  return status;"
 
             # Complete
@@ -276,10 +280,6 @@ def wrapper_cblas(routine):
                             extra_argument += "," + NL + indent
                             extra_argument += "reinterpret_cast<return_pointer_" + flavour.buffer_type[:-1] + ">"
                             extra_argument += "(&" + output_buffer + "_buffer[" + output_buffer + "_offset])"
-                        elif output_buffer in routine.index_buffers():
-                            assignment = "reinterpret_cast<int*>(&" + output_buffer + "_buffer[0])[" + output_buffer + "_offset] = static_cast<int>("
-                            postpostfix = ")"
-                            indent += " " * (len(assignment) + 1)
                         else:
                             assignment = output_buffer + "_buffer[" + output_buffer + "_offset]"
                             if flavour.name in ["Sc", "Dz"]:
@@ -299,7 +299,10 @@ def wrapper_cblas(routine):
 
                 # Convert to float (note: also integer buffers are stored as half/float)
                 for buf in routine.inputs + routine.outputs:
-                    result += "  auto " + buf + "_buffer_bis = HalfToFloatBuffer(" + buf + "_buffer);" + NL
+                    if buf not in routine.index_buffers():
+                        result += "  auto " + buf + "_buffer_bis = HalfToFloatBuffer(" + buf + "_buffer);" + NL
+                    else:
+                        result += "  auto " + buf + "_buffer_bis = " + buf + "_buffer;" + NL
 
                 # Call the float routine
                 result += "  cblasX" + routine.name + "("
@@ -308,7 +311,8 @@ def wrapper_cblas(routine):
 
                 # Convert back to half
                 for buf in routine.outputs:
-                    result += "  FloatToHalfBuffer(" + buf + "_buffer, " + buf + "_buffer_bis);" + NL
+                    if buf not in routine.index_buffers():
+                        result += "  FloatToHalfBuffer(" + buf + "_buffer, " + buf + "_buffer_bis);" + NL
 
             # Complete
             result += "}" + NL
diff --git a/scripts/generator/generator/routine.py b/scripts/generator/generator/routine.py
index 8b6ab57f..c2201c0d 100644
--- a/scripts/generator/generator/routine.py
+++ b/scripts/generator/generator/routine.py
@@ -282,7 +282,10 @@ class Routine:
         """As above but for OpenCL"""
         prefix = "const " if name in self.inputs else ""
         if name in self.inputs or name in self.outputs:
-            a = [prefix + "Buffer<" + flavour.buffer_type + ">& " + name + "_buffer"]
+            if name == "imax":
+                a = [prefix + "Buffer<unsigned int>& " + name + "_buffer"]
+            else:
+                a = [prefix + "Buffer<" + flavour.buffer_type + ">& " + name + "_buffer"]
             b = ["const size_t " + name + "_offset"]
             c = ["const size_t " + name + "_" + self.postfix(name)] if name not in self.buffers_without_ld_inc() else []
             return [", ".join(a + b + c)]
@@ -292,7 +295,10 @@ class Routine:
         """As above but for CUDA"""
         prefix = "const " if name in self.inputs else ""
         if name in self.inputs or name in self.outputs:
-            a = [prefix + flavour.buffer_type + "* " + name + "_buffer"]
+            if name == "imax":
+                a = [prefix + "unsigned int * " + name + "_buffer"]
+            else:
+                a = [prefix + flavour.buffer_type + "* " + name + "_buffer"]
             b = ["const size_t " + name + "_offset"]
             c = ["const size_t " + name + "_" + self.postfix(name)] if name not in self.buffers_without_ld_inc() else []
             return [", ".join(a + b + c)]
@@ -302,7 +308,10 @@ class Routine:
         """As above but as vectors"""
         prefix = "const " if name in self.inputs else ""
         if name in self.inputs or name in self.outputs:
-            a = [prefix + "std::vector<" + flavour.buffer_type + ">& " + name + "_buffer"]
+            if name == "imax":
+                a = [prefix + "std::vector<unsigned int>& " + name + "_buffer"]
+            else:
+                a = [prefix + "std::vector<" + flavour.buffer_type + ">& " + name + "_buffer"]
             b = ["const size_t " + name + "_offset"]
             c = ["const size_t " + name + "_" + self.postfix(name)] if name not in self.buffers_without_ld_inc() else []
             return [", ".join(a + b + c)]