[MTAI-484] feat(build): fix code style for cpp lint

hanhaowen-mt · Aug 16, 2023 · ce74631 · ce74631
1 parent 2a7c365
commit ce74631
Show file tree

Hide file tree

Showing 10 changed files with 19 additions and 17 deletions.
diff --git a/paddle/phi/backends/CMakeLists.txt b/paddle/phi/backends/CMakeLists.txt
@@ -7,7 +7,9 @@ if(NOT APPLE AND NOT WIN32)
   list(APPEND BACKENDS_SRCS device_code.cc)
 endif()
 
-if(WITH_GPU OR WITH_ROCM OR WITH_MUSA)
+if(WITH_GPU
+   OR WITH_ROCM
+   OR WITH_MUSA)
   list(APPEND BACKENDS_SRCS gpu/gpu_context.cc gpu/gpu_info.cc
        gpu/gpu_resources.cc)
   if(WITH_GPU)

diff --git a/paddle/phi/backends/dynload/mublas.cc b/paddle/phi/backends/dynload/mublas.cc
@@ -25,4 +25,3 @@ MUBLAS_BLAS_ROUTINE_EACH(DEFINE_WRAP);
 
 }  // namespace dynload
 }  // namespace phi
-
diff --git a/paddle/phi/backends/dynload/musa_driver.h b/paddle/phi/backends/dynload/musa_driver.h
@@ -61,7 +61,6 @@ extern bool HasCUDADriver();
   __macro(muDeviceGetAttribute);                        \
   __macro(muDeviceGet);
 
-
 MUSA_ROUTINE_EACH(DECLARE_DYNAMIC_LOAD_MUSA_WRAP);
 
 #undef DECLARE_DYNAMIC_LOAD_MUSA_WRAP

diff --git a/paddle/phi/common/cpstring_impl.h b/paddle/phi/common/cpstring_impl.h
@@ -77,7 +77,8 @@ HOSTDEVICE static inline uint32_t swap32(uint32_t host_int) {
 }
 #endif
 
-#if PD_PSTRING_LITTLE_ENDIAN || (defined(__NVCC__) || defined(__HIPCC__) || defined(__MUSACC__))
+#if PD_PSTRING_LITTLE_ENDIAN || \
+    (defined(__NVCC__) || defined(__HIPCC__) || defined(__MUSACC__))
 #define PD_le32toh(x) x
 #else  // PD_PSTRING_LITTLE_ENDIAN
 #define PD_le32toh(x) swap32(x)

diff --git a/paddle/phi/core/hostdevice.h b/paddle/phi/core/hostdevice.h
@@ -30,7 +30,8 @@
 #include "xpu/kernel/math.h"
 #endif
 
-#if (defined(__CUDACC__) || defined(__HIPCC__) || defined(__MUSACC__) || defined(__xpu__))
+#if (defined(__CUDACC__) || defined(__HIPCC__) || defined(__MUSACC__) || \
+     defined(__xpu__))
 #define HOSTDEVICE __host__ __device__
 #define DEVICE __device__
 #define HOST __host__

diff --git a/paddle/phi/kernels/funcs/embedding_grad.h b/paddle/phi/kernels/funcs/embedding_grad.h
@@ -96,9 +96,9 @@ __global__ void EmbeddingGradDeterministicKernel(T* table,
         unsigned long long int matchmask =      // NOLINT
             __ballot(match_found_this_thread);  // NOLINT
         int first_remaining_peer = __ffsll(matchmask) - 1;
-#else // MUSA and CUDA
-        // If and only if match_found_this_thread of the Nth thread is non-zero,
-        // set the Nth bit of matchmask to 1.
+#else  // MUSA and CUDA
+       // If and only if match_found_this_thread of the Nth thread is non-zero,
+       // set the Nth bit of matchmask to 1.
         unsigned int matchmask =
             __ballot_sync(0xffffffff, match_found_this_thread);
         // Find the position of the first bit set to 1 in matchmask.
@@ -112,7 +112,7 @@ __global__ void EmbeddingGradDeterministicKernel(T* table,
           while (matchmask) {
 #ifdef PADDLE_WITH_HIP
             first_remaining_peer = __ffsll(matchmask) - 1;
-#else // CUDA and MUSA
+#else  // CUDA and MUSA
             first_remaining_peer = __ffs(matchmask) - 1;
 #endif
             my_s[threadIdx.x] +=
@@ -142,7 +142,7 @@ void LaunchEmbeddingGradDeterministicKernel(const GPUContext& ctx,
 #ifdef PADDLE_WITH_HIP
   constexpr int kWarpSize = 64;
   constexpr int kBlockDimY = 16;
-#else // CUDA and MUSA
+#else  // CUDA and MUSA
   constexpr int kWarpSize = 32;
   constexpr int kBlockDimY = 32;
 #endif

diff --git a/paddle/phi/kernels/funcs/layer_norm_impl.cu.h b/paddle/phi/kernels/funcs/layer_norm_impl.cu.h
@@ -1353,8 +1353,8 @@ __global__ void LayerNormBackwardComputeGradInput(const T *__restrict__ dout,
       // WARP_SHFL_XOR(sum_loss, mask);
       sum_loss1 += __shfl_xor(sum_loss1, mask, warpSize);
       sum_loss2 += __shfl_xor(sum_loss2, mask, warpSize);
-#else // CUDA and MUSA
-      // WARP_SHFL_XOR(sum_loss, mask);
+#else  // CUDA and MUSA
+       // WARP_SHFL_XOR(sum_loss, mask);
       sum_loss1 += __shfl_xor_sync(0xffffffff, sum_loss1, mask, warpSize);
       sum_loss2 += __shfl_xor_sync(0xffffffff, sum_loss2, mask, warpSize);
 #endif
@@ -1504,8 +1504,8 @@ __global__ void LayerNormBackwardComputeGradInputWithSmallFeatureSize(
       // WARP_SHFL_XOR(sum_loss, mask);
       sum_loss1 += __shfl_xor(sum_loss1, mask, warpSize);
       sum_loss2 += __shfl_xor(sum_loss2, mask, warpSize);
-#else // CUDA and MUSA
-      // WARP_SHFL_XOR(sum_loss, mask);
+#else  // CUDA and MUSA
+       // WARP_SHFL_XOR(sum_loss, mask);
       sum_loss1 += __shfl_xor_sync(0xffffffff, sum_loss1, mask, WarpSize);
       sum_loss2 += __shfl_xor_sync(0xffffffff, sum_loss2, mask, WarpSize);
 #endif

diff --git a/paddle/phi/kernels/gpu/check_numerics_kernel.cu b/paddle/phi/kernels/gpu/check_numerics_kernel.cu
@@ -503,7 +503,7 @@ void CheckNumericsKernel(const Context& ctx,
 #ifdef PADDLE_WITH_MUSA
   PADDLE_THROW(phi::errors::Unimplemented(
       "OP check_numerics is unsupported for MUSA backend now!"));
-return;
+  return;
 #else
   int dev_id = tensor.place().device;
   VLOG(6) << "op_type=" << op_type << ", var_name=" << var_name

diff --git a/paddle/phi/kernels/impl/elementwise_kernel_impl.h b/paddle/phi/kernels/impl/elementwise_kernel_impl.h
@@ -17,7 +17,8 @@
 #include "paddle/phi/kernels/elementwise_kernel.h"
 #include "paddle/phi/kernels/funcs/elementwise_base.h"
 #include "paddle/phi/kernels/funcs/elementwise_functor.h"
-#if defined(__NVCC__) || defined(__HIPCC__) || defined(__MUSACC__) || defined(__xpu__)
+#if defined(__NVCC__) || defined(__HIPCC__) || defined(__MUSACC__) || \
+    defined(__xpu__)
 #include "paddle/phi/kernels/funcs/broadcast_function.h"
 #endif
 

diff --git a/paddle/phi/kernels/impl/matmul_kernel_impl.h b/paddle/phi/kernels/impl/matmul_kernel_impl.h
@@ -1006,7 +1006,6 @@ void MatmulWithFlattenKernel(const Context& dev_ctx,
   }
 
   auto blas = phi::funcs::GetBlas<Context, T>(dev_ctx);
-
   blas.MatMul(x_matrix, y_matrix, out);
   if (z_dim.size() != 2) {
     out->Resize(z_dim);
Original file line number	Diff line number	Diff line change
Expand Up		@@ -25,4 +25,3 @@ MUBLAS_BLAS_ROUTINE_EACH(DEFINE_WRAP);

		} // namespace dynload
		} // namespace phi