[kernels] remove GKO_DEVICE_NAMESPACE

ginkgo-project · Aug 20, 2024 · b35d079 · b35d079
1 parent 590713c
commit b35d079
Show file tree

Hide file tree

Showing 3 changed files with 30 additions and 39 deletions.
diff --git a/common/cuda_hip/base/batch_multi_vector_kernels.cpp b/common/cuda_hip/base/batch_multi_vector_kernels.cpp
@@ -37,19 +37,19 @@ void scale(std::shared_ptr<const DefaultExecutor> exec,
     const auto alpha_ub = get_batch_struct(alpha);
     const auto x_ub = get_batch_struct(x);
     if (alpha->get_common_size()[1] == 1) {
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::scale_kernel<<<
-            num_blocks, default_block_size, 0, exec->get_stream()>>>(
+        batch_single_kernels::scale_kernel<<<num_blocks, default_block_size, 0,
+                                             exec->get_stream()>>>(
             alpha_ub, x_ub,
             [] __device__(int row, int col, int stride) { return 0; });
     } else if (alpha->get_common_size() == x->get_common_size()) {
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::scale_kernel<<<
-            num_blocks, default_block_size, 0, exec->get_stream()>>>(
+        batch_single_kernels::scale_kernel<<<num_blocks, default_block_size, 0,
+                                             exec->get_stream()>>>(
             alpha_ub, x_ub, [] __device__(int row, int col, int stride) {
                 return row * stride + col;
             });
     } else {
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::scale_kernel<<<
-            num_blocks, default_block_size, 0, exec->get_stream()>>>(
+        batch_single_kernels::scale_kernel<<<num_blocks, default_block_size, 0,
+                                             exec->get_stream()>>>(
             alpha_ub, x_ub,
             [] __device__(int row, int col, int stride) { return col; });
     }
@@ -71,11 +71,11 @@ void add_scaled(std::shared_ptr<const DefaultExecutor> exec,
     const auto x_ub = get_batch_struct(x);
     const auto y_ub = get_batch_struct(y);
     if (alpha->get_common_size()[1] == 1) {
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::add_scaled_kernel<<<
+        batch_single_kernels::add_scaled_kernel<<<
             num_blocks, default_block_size, 0, exec->get_stream()>>>(
             alpha_ub, x_ub, y_ub, [] __device__(int col) { return 0; });
     } else {
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::add_scaled_kernel<<<
+        batch_single_kernels::add_scaled_kernel<<<
             num_blocks, default_block_size, 0, exec->get_stream()>>>(
             alpha_ub, x_ub, y_ub, [] __device__(int col) { return col; });
     }
@@ -96,10 +96,9 @@ void compute_dot(std::shared_ptr<const DefaultExecutor> exec,
     const auto x_ub = get_batch_struct(x);
     const auto y_ub = get_batch_struct(y);
     const auto res_ub = get_batch_struct(result);
-    GKO_DEVICE_NAMESPACE::batch_single_kernels::
-        compute_gen_dot_product_kernel<<<num_blocks, default_block_size, 0,
-                                         exec->get_stream()>>>(
-            x_ub, y_ub, res_ub, [] __device__(auto val) { return val; });
+    batch_single_kernels::compute_gen_dot_product_kernel<<<
+        num_blocks, default_block_size, 0, exec->get_stream()>>>(
+        x_ub, y_ub, res_ub, [] __device__(auto val) { return val; });
 }
 
 GKO_INSTANTIATE_FOR_EACH_VALUE_TYPE(
@@ -117,10 +116,9 @@ void compute_conj_dot(std::shared_ptr<const DefaultExecutor> exec,
     const auto x_ub = get_batch_struct(x);
     const auto y_ub = get_batch_struct(y);
     const auto res_ub = get_batch_struct(result);
-    GKO_DEVICE_NAMESPACE::batch_single_kernels::
-        compute_gen_dot_product_kernel<<<num_blocks, default_block_size, 0,
-                                         exec->get_stream()>>>(
-            x_ub, y_ub, res_ub, [] __device__(auto val) { return conj(val); });
+    batch_single_kernels::compute_gen_dot_product_kernel<<<
+        num_blocks, default_block_size, 0, exec->get_stream()>>>(
+        x_ub, y_ub, res_ub, [] __device__(auto val) { return conj(val); });
 }
 
 GKO_INSTANTIATE_FOR_EACH_VALUE_TYPE(
@@ -136,8 +134,9 @@ void compute_norm2(std::shared_ptr<const DefaultExecutor> exec,
     const auto num_rhs = x->get_common_size()[1];
     const auto x_ub = get_batch_struct(x);
     const auto res_ub = get_batch_struct(result);
-    GKO_DEVICE_NAMESPACE::batch_single_kernels::compute_norm2_kernel<<<
-        num_blocks, default_block_size, 0, exec->get_stream()>>>(x_ub, res_ub);
+    batch_single_kernels::compute_norm2_kernel<<<num_blocks, default_block_size,
+                                                 0, exec->get_stream()>>>(
+        x_ub, res_ub);
 }
 
 GKO_INSTANTIATE_FOR_EACH_VALUE_TYPE(
@@ -152,7 +151,7 @@ void copy(std::shared_ptr<const DefaultExecutor> exec,
     const auto num_blocks = x->get_num_batch_items();
     const auto result_ub = get_batch_struct(result);
     const auto x_ub = get_batch_struct(x);
-    GKO_DEVICE_NAMESPACE::batch_single_kernels::
+    batch_single_kernels::
         copy_kernel<<<num_blocks, default_block_size, 0, exec->get_stream()>>>(
             x_ub, result_ub);
 }

diff --git a/omp/base/batch_multi_vector_kernels.cpp b/omp/base/batch_multi_vector_kernels.cpp
@@ -33,7 +33,7 @@ void scale(std::shared_ptr<const DefaultExecutor> exec,
     for (size_type batch = 0; batch < x->get_num_batch_items(); ++batch) {
         const auto alpha_b = gko::batch::extract_batch_item(alpha_ub, batch);
         const auto x_b = gko::batch::extract_batch_item(x_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::scale_kernel(alpha_b, x_b);
+        batch_single_kernels::scale_kernel(alpha_b, x_b);
     }
 }
 
@@ -55,8 +55,7 @@ void add_scaled(std::shared_ptr<const DefaultExecutor> exec,
         const auto alpha_b = gko::batch::extract_batch_item(alpha_ub, batch);
         const auto x_b = gko::batch::extract_batch_item(x_ub, batch);
         const auto y_b = gko::batch::extract_batch_item(y_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::add_scaled_kernel(alpha_b,
-                                                                      x_b, y_b);
+        batch_single_kernels::add_scaled_kernel(alpha_b, x_b, y_b);
     }
 }
 
@@ -78,8 +77,7 @@ void compute_dot(std::shared_ptr<const DefaultExecutor> exec,
         const auto res_b = gko::batch::extract_batch_item(res_ub, batch);
         const auto x_b = gko::batch::extract_batch_item(x_ub, batch);
         const auto y_b = gko::batch::extract_batch_item(y_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::compute_dot_product_kernel(
-            x_b, y_b, res_b);
+        batch_single_kernels::compute_dot_product_kernel(x_b, y_b, res_b);
     }
 }
 
@@ -101,8 +99,7 @@ void compute_conj_dot(std::shared_ptr<const DefaultExecutor> exec,
         const auto res_b = gko::batch::extract_batch_item(res_ub, batch);
         const auto x_b = gko::batch::extract_batch_item(x_ub, batch);
         const auto y_b = gko::batch::extract_batch_item(y_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::
-            compute_conj_dot_product_kernel(x_b, y_b, res_b);
+        batch_single_kernels::compute_conj_dot_product_kernel(x_b, y_b, res_b);
     }
 }
 
@@ -121,8 +118,7 @@ void compute_norm2(std::shared_ptr<const DefaultExecutor> exec,
     for (size_type batch = 0; batch < result->get_num_batch_items(); ++batch) {
         const auto res_b = gko::batch::extract_batch_item(res_ub, batch);
         const auto x_b = gko::batch::extract_batch_item(x_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::compute_norm2_kernel(x_b,
-                                                                         res_b);
+        batch_single_kernels::compute_norm2_kernel(x_b, res_b);
     }
 }
 
@@ -141,7 +137,7 @@ void copy(std::shared_ptr<const DefaultExecutor> exec,
     for (size_type batch = 0; batch < x->get_num_batch_items(); ++batch) {
         const auto result_b = gko::batch::extract_batch_item(result_ub, batch);
         const auto x_b = gko::batch::extract_batch_item(x_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::copy_kernel(x_b, result_b);
+        batch_single_kernels::copy_kernel(x_b, result_b);
     }
 }
 

diff --git a/reference/base/batch_multi_vector_kernels.cpp b/reference/base/batch_multi_vector_kernels.cpp
@@ -35,7 +35,7 @@ void scale(std::shared_ptr<const DefaultExecutor> exec,
     for (size_type batch = 0; batch < x->get_num_batch_items(); ++batch) {
         const auto alpha_b = batch::extract_batch_item(alpha_ub, batch);
         const auto x_b = batch::extract_batch_item(x_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::scale_kernel(alpha_b, x_b);
+        batch_single_kernels::scale_kernel(alpha_b, x_b);
     }
 }
 
@@ -56,8 +56,7 @@ void add_scaled(std::shared_ptr<const DefaultExecutor> exec,
         const auto alpha_b = batch::extract_batch_item(alpha_ub, batch);
         const auto x_b = batch::extract_batch_item(x_ub, batch);
         const auto y_b = batch::extract_batch_item(y_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::add_scaled_kernel(alpha_b,
-                                                                      x_b, y_b);
+        batch_single_kernels::add_scaled_kernel(alpha_b, x_b, y_b);
     }
 }
 
@@ -78,8 +77,7 @@ void compute_dot(std::shared_ptr<const DefaultExecutor> exec,
         const auto res_b = batch::extract_batch_item(res_ub, batch);
         const auto x_b = batch::extract_batch_item(x_ub, batch);
         const auto y_b = batch::extract_batch_item(y_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::compute_dot_product_kernel(
-            x_b, y_b, res_b);
+        batch_single_kernels::compute_dot_product_kernel(x_b, y_b, res_b);
     }
 }
 
@@ -100,8 +98,7 @@ void compute_conj_dot(std::shared_ptr<const DefaultExecutor> exec,
         const auto res_b = batch::extract_batch_item(res_ub, batch);
         const auto x_b = batch::extract_batch_item(x_ub, batch);
         const auto y_b = batch::extract_batch_item(y_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::
-            compute_conj_dot_product_kernel(x_b, y_b, res_b);
+        batch_single_kernels::compute_conj_dot_product_kernel(x_b, y_b, res_b);
     }
 }
 
@@ -119,8 +116,7 @@ void compute_norm2(std::shared_ptr<const DefaultExecutor> exec,
     for (size_type batch = 0; batch < result->get_num_batch_items(); ++batch) {
         const auto res_b = batch::extract_batch_item(res_ub, batch);
         const auto x_b = batch::extract_batch_item(x_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::compute_norm2_kernel(x_b,
-                                                                         res_b);
+        batch_single_kernels::compute_norm2_kernel(x_b, res_b);
     }
 }
 
@@ -138,7 +134,7 @@ void copy(std::shared_ptr<const DefaultExecutor> exec,
     for (size_type batch = 0; batch < x->get_num_batch_items(); ++batch) {
         const auto result_b = batch::extract_batch_item(result_ub, batch);
         const auto x_b = batch::extract_batch_item(x_ub, batch);
-        GKO_DEVICE_NAMESPACE::batch_single_kernels::copy_kernel(x_b, result_b);
+        batch_single_kernels::copy_kernel(x_b, result_b);
     }
 }