PaddlePaddle
diff --git a/‎paddle/phi/kernels/CMakeLists.txt
Copy file name to clipboard
+18-31Lines changed: 18 additions & 31 deletions b/‎paddle/phi/kernels/CMakeLists.txt
Copy file name to clipboard
+18-31Lines changed: 18 additions & 31 deletions
diff --git a/‎paddle/phi/kernels/cal_aux_loss_grad_kernel.h
Copy file name to clipboard
-31Lines changed: 0 additions & 31 deletions b/‎paddle/phi/kernels/cal_aux_loss_grad_kernel.h
Copy file name to clipboard
-31Lines changed: 0 additions & 31 deletions
diff --git a/‎paddle/phi/kernels/cal_aux_loss_kernel.h
Copy file name to clipboard
-35Lines changed: 0 additions & 35 deletions b/‎paddle/phi/kernels/cal_aux_loss_kernel.h
Copy file name to clipboard
-35Lines changed: 0 additions & 35 deletions
diff --git a/‎paddle/phi/kernels/expand_modality_expert_id_kernel.h
Copy file name to clipboard
-29Lines changed: 0 additions & 29 deletions b/‎paddle/phi/kernels/expand_modality_expert_id_kernel.h
Copy file name to clipboard
-29Lines changed: 0 additions & 29 deletions
diff --git a/‎paddle/phi/kernels/int_bincount.h
Copy file name to clipboard
-28Lines changed: 0 additions & 28 deletions b/‎paddle/phi/kernels/int_bincount.h
Copy file name to clipboard
-28Lines changed: 0 additions & 28 deletions
diff --git a/‎paddle/phi/kernels/gpu/cal_aux_loss_grad_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/cal_aux_loss_grad_kernel.cu
Copy file name to clipboard
+1-3Lines changed: 1 addition & 3 deletions b/‎paddle/phi/kernels/gpu/cal_aux_loss_grad_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/cal_aux_loss_grad_kernel.cu
Copy file name to clipboard
+1-3Lines changed: 1 addition & 3 deletions
diff --git a/‎paddle/phi/kernels/gpu/cal_aux_loss_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/cal_aux_loss_kernel.cu
Copy file name to clipboard
+1-4Lines changed: 1 addition & 4 deletions b/‎paddle/phi/kernels/gpu/cal_aux_loss_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/cal_aux_loss_kernel.cu
Copy file name to clipboard
+1-4Lines changed: 1 addition & 4 deletions
diff --git a/‎paddle/phi/kernels/gpu/expand_modality_expert_id_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/expand_modality_expert_id_kernel.cu
Copy file name to clipboard
+1-2Lines changed: 1 addition & 2 deletions b/‎paddle/phi/kernels/gpu/expand_modality_expert_id_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/expand_modality_expert_id_kernel.cu
Copy file name to clipboard
+1-2Lines changed: 1 addition & 2 deletions
diff --git a/‎paddle/phi/kernels/gpu/build_src_rank_and_local_expert_id_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/ext_build_src_rank_and_local_expert_id_kernel.cu
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion b/‎paddle/phi/kernels/gpu/build_src_rank_and_local_expert_id_kernel.cu renamed to ‎paddle/phi/kernels/legacy/gpu/ext_build_src_rank_and_local_expert_id_kernel.cu
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion
diff --git a/‎paddle/phi/kernels/gpu/int_bincount.cu renamed to ‎paddle/phi/kernels/legacy/gpu/int_bincount.cu
Copy file name to clipboard
+1-2Lines changed: 1 addition & 2 deletions b/‎paddle/phi/kernels/gpu/int_bincount.cu renamed to ‎paddle/phi/kernels/legacy/gpu/int_bincount.cu
Copy file name to clipboard
+1-2Lines changed: 1 addition & 2 deletions
@@ -52,24 +52,25 @@ endif()
 # New Op only supported by CUDA>=12.0 and Linux
 if(((WITH_GPU) AND (CUDA_VERSION VERSION_LESS 12.0))
    OR APPLE
-   OR WIN32)
+   OR WIN32
+   OR WITH_ROCM)
   list(
     REMOVE_ITEM
-    kernel_gpu
-    "gpu/moe_ops_partial_nosoftmaxtopk_kernel.cu"
-    "gpu/moe_ops_partial_nosoftmaxtopk_grad_kernel.cu"
-    "gpu/moe_gate_dispatch_permute_grad_kernel.cu"
-    "gpu/moe_gate_dispatch_permute_kernel.cu"
-    "gpu/expand_modality_expert_id_kernel.cu"
-    "gpu/moe_combine_kernel.cu"
-    "gpu/moe_combine_grad_kernel.cu"
-    "gpu/cal_aux_loss_kernel.cu"
-    "gpu/cal_aux_loss_grad_kernel.cu"
-    "gpu/build_src_rank_and_local_expert_id_kernel.cu"
-    "gpu/moe_gate_dispatch_kernel.cu"
-    "gpu/moe_gate_dispatch_grad_kernel.cu"
-    "gpu/int_bincount.cu"
-    "gpu/layer_norm_cuda_kernel.cu")
+    kernel_cu
+    "legacy/gpu/moe_ops_partial_nosoftmaxtopk_kernel.cu"
+    "legacy/gpu/moe_ops_partial_nosoftmaxtopk_grad_kernel.cu"
+    "legacy/gpu/moe_gate_dispatch_permute_grad_kernel.cu"
+    "legacy/gpu/moe_gate_dispatch_permute_kernel.cu"
+    "legacy/gpu/expand_modality_expert_id_kernel.cu"
+    "legacy/gpu/moe_combine_kernel.cu"
+    "legacy/gpu/moe_combine_grad_kernel.cu"
+    "legacy/gpu/cal_aux_loss_kernel.cu"
+    "legacy/gpu/cal_aux_loss_grad_kernel.cu"
+    "legacy/gpu/ext_build_src_rank_and_local_expert_id_kernel.cu"
+    "legacy/gpu/moe_gate_dispatch_kernel.cu"
+    "legacy/gpu/moe_gate_dispatch_grad_kernel.cu"
+    "legacy/gpu/int_bincount.cu"
+    "legacy/gpu/layer_norm_cuda_kernel.cu")
 endif()
 
 if(NOT WITH_DGC)
@@ -251,29 +252,15 @@ if(WITH_ROCM)
   list(
     REMOVE_ITEM
     kernel_gpu
-    "gpu/moe_ops_partial_nosoftmaxtopk_kernel.cu"
-    "gpu/moe_ops_partial_nosoftmaxtopk_grad_kernel.cu"
-    "gpu/moe_gate_dispatch_permute_grad_kernel.cu"
-    "gpu/moe_gate_dispatch_permute_kernel.cu"
-    "gpu/expand_modality_expert_id_kernel.cu"
-    "gpu/moe_combine_kernel.cu"
-    "gpu/moe_combine_grad_kernel.cu"
     "gpu/affine_grid_grad_kernel.cu"
-    "gpu/cal_aux_loss_kernel.cu"
-    "gpu/cal_aux_loss_grad_kernel.cu"
-    "gpu/build_src_rank_and_local_expert_id_kernel.cu"
-    "gpu/moe_gate_dispatch_kernel.cu"
-    "gpu/moe_gate_dispatch_grad_kernel.cu"
     "gpu/apply_per_channel_scale_kernel.cu"
     "gpu/calc_reduced_attn_kernel.cu"
     "gpu/eigvalsh_kernel.cu"
     "gpu/lu_kernel.cu"
     "gpu/matrix_rank_kernel.cu"
     "gpu/matrix_rank_tol_kernel.cu"
     "gpu/svd_kernel.cu"
-    "gpu/cuda_gemm_kernel.cu"
-    "gpu/int_bincount.cu"
-    "gpu/layer_norm_cuda_kernel.cu")
+    "gpu/cuda_gemm_kernel.cu")
 endif()
 
 # Remove AP kernel when CINN is not enabled.
 
@@ -12,9 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#pragma once
-
-#include "paddle/phi/kernels/cal_aux_loss_grad_kernel.h"
+#include "paddle/phi/core/dense_tensor.h"
 
 #include "paddle/phi/backends/gpu/gpu_context.h"
 #include "paddle/phi/core/kernel_registry.h"
 
@@ -12,11 +12,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#pragma once
-
-#include "paddle/phi/kernels/cal_aux_loss_kernel.h"
-
 #include "paddle/phi/backends/gpu/gpu_context.h"
+#include "paddle/phi/core/dense_tensor.h"
 #include "paddle/phi/core/kernel_registry.h"
 
 #include "paddle/phi/kernels/funcs/math_cuda_utils.h"
 
@@ -11,10 +11,9 @@
 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 // See the License for the specific language governing permissions and
 // limitations under the License.
-
-#include "paddle/phi/kernels/expand_modality_expert_id_kernel.h"
 #include <thrust/device_vector.h>
 #include "paddle/phi/backends/gpu/gpu_context.h"
+#include "paddle/phi/core/dense_tensor.h"
 #include "paddle/phi/core/kernel_registry.h"
 
 namespace phi {
 
@@ -1,4 +1,4 @@
-// Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
+// Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
 
@@ -1,4 +1,3 @@
-// NOLINT
 // Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
@@ -14,11 +13,11 @@
 // limitations under the License.
 
 // #include "paddle/extension.h"
-#include "paddle/phi/kernels/int_bincount.h"  // NOLINT
 #include <cstdint>
 #include <vector>
 #include "cub/device/device_histogram.cuh"
 #include "paddle/common/flags.h"
+#include "paddle/phi/core/dense_tensor.h"
 #include "paddle/phi/core/utils/data_type.h"
 #include "paddle/phi/kernels/empty_kernel.h"  // NOLINT
Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,4 @@`
`1`		`-// Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.`
	`1`	`+// Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.`
`2`	`2`	`//`
`3`	`3`	`// Licensed under the Apache License, Version 2.0 (the "License");`
`4`	`4`	`// you may not use this file except in compliance with the License.`