remove .hip extension

src/include/gridwise_convolution_implicit_gemm_v1r3_nchw_cyxk_nkhw.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v1r3_nchw_cyxk_nkhw.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_2d_tensor_op.hip.hpp"
-#include "blockwise_tensor_slice_op.hip.hpp"
-#include "threadwise_tensor_slice_op.hip.hpp"
-#include "threadwise_4d_tensor_op.hip.hpp"
-#include "blockwise_batched_gemm.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_2d_tensor_op.hpp"
+#include "blockwise_tensor_slice_op.hpp"
+#include "threadwise_tensor_slice_op.hpp"
+#include "threadwise_4d_tensor_op.hpp"
+#include "blockwise_batched_gemm.hpp"
 
 template <index_t GridSize,
           index_t BlockSize,

src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_4d_tensor_op.hip.hpp"
-#include "blockwise_2d_tensor_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_4d_tensor_op.hpp"
+#include "blockwise_2d_tensor_op.hpp"
+#include "blockwise_gemm.hpp"
 
 // define B = flatten(N, Hi, Wi)
 template <index_t GridSize,

src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn_lds_double_buffer.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn_lds_double_buffer.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_4d_tensor_op.hip.hpp"
-#include "blockwise_2d_tensor_op.hip.hpp"
-#include "threadwise_tensor_slice_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_4d_tensor_op.hpp"
+#include "blockwise_2d_tensor_op.hpp"
+#include "threadwise_tensor_slice_op.hpp"
+#include "blockwise_gemm.hpp"
 
 // define B = flatten(N, Hi, Wi)
 template <index_t GridSize,

src/include/gridwise_convolution_implicit_gemm_v3_lds_double_buffer_nchw_cyxk_nkhw.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v3_lds_double_buffer_nchw_cyxk_nkhw.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMergedTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_generic_tensor_slice_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMergedTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_generic_tensor_slice_op.hpp"
+#include "blockwise_gemm.hpp"
 
 // define B = merge(N0, Ho, Wo)
 template <index_t GridSize,

src/include/gridwise_convolution_implicit_gemm_v3_nchw_cyxk_nkhw.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v3_nchw_cyxk_nkhw.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMergedTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_generic_tensor_slice_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMergedTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_generic_tensor_slice_op.hpp"
+#include "blockwise_gemm.hpp"
 
 // define B = merge(N0, Ho, Wo)
 template <index_t GridSize,

src/include/gridwise_convolution_implicit_gemm_v4_lds_double_buffer_nchw_kcyx_nkhw.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v4_lds_double_buffer_nchw_kcyx_nkhw.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMergedTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_generic_tensor_slice_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
-#include "threadwise_generic_tensor_slice_op.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMergedTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_generic_tensor_slice_op.hpp"
+#include "blockwise_gemm.hpp"
+#include "threadwise_generic_tensor_slice_op.hpp"
 
 // define B = merge(N0, Ho, Wo)
 template <index_t GridSize,

src/include/gridwise_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hip.hpp → src/include/gridwise_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMergedTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_generic_tensor_slice_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
-#include "threadwise_generic_tensor_slice_op.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMergedTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_generic_tensor_slice_op.hpp"
+#include "blockwise_gemm.hpp"
+#include "threadwise_generic_tensor_slice_op.hpp"
 
 // define B = merge(N0, Ho, Wo)
 template <index_t GridSize,

src/include/gridwise_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hip.hpp → src/include/gridwise_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "blockwise_2d_tensor_op.hip.hpp"
-#include "blockwise_4d_tensor_op.hip.hpp"
-#include "blockwise_direct_convolution.hip.hpp"
-#include "threadwise_4d_tensor_op.hip.hpp"
-#include "threadwise_direct_convolution.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "blockwise_2d_tensor_op.hpp"
+#include "blockwise_4d_tensor_op.hpp"
+#include "blockwise_direct_convolution.hpp"
+#include "threadwise_4d_tensor_op.hpp"
+#include "threadwise_direct_convolution.hpp"
 
 template <class TInWei,
           class TOut,

src/include/gridwise_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hip.hpp → src/include/gridwise_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
-#include "blockwise_4d_tensor_op.hip.hpp"
-#include "blockwise_2d_tensor_op.hip.hpp"
-#include "threadwise_4d_tensor_op.hip.hpp"
-#include "blockwise_gemm.hip.hpp"
+#include "common.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMatrixDescriptor.hpp"
+#include "blockwise_4d_tensor_op.hpp"
+#include "blockwise_2d_tensor_op.hpp"
+#include "threadwise_4d_tensor_op.hpp"
+#include "blockwise_gemm.hpp"
 
 template <index_t GridSize,
           index_t BlockSize,

src/include/threadwise_4d_tensor_op.hip.hpp → src/include/threadwise_4d_tensor_op.hpp

 #pragma once
-#include "ConstantTensorDescriptor.hip.hpp"
+#include "ConstantTensorDescriptor.hpp"
 
 template <class Float, class Desc, class IDim, class NShift>
 __device__ void threadwise_4d_tensor_shift_down(Desc, Float* __restrict__ p, IDim, NShift)

src/include/threadwise_direct_convolution.hip.hpp → src/include/threadwise_direct_convolution.hpp

 #pragma once
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "threadwise_tensor_slice_op.hip.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "threadwise_tensor_slice_op.hpp"
 
 // optimized for scenario if p_in, p_wei, p_out are in register
 template <class TInWei, class TOut, class InDesc, class WeiDesc, class OutDesc>

src/include/threadwise_gemm.hip.hpp → src/include/threadwise_gemm.hpp

 #pragma once
-#include "common.hip.hpp"
-#include "ConstantMatrixDescriptor.hip.hpp"
+#include "common.hpp"
+#include "ConstantMatrixDescriptor.hpp"
 
 template <class Float, class Matrix>
 __device__ void threadwise_matrix_set_zero(Matrix, Float* __restrict__ p_thread)

src/include/threadwise_generic_tensor_slice_op.hip.hpp → src/include/threadwise_generic_tensor_slice_op.hpp

 #pragma once
-#include "ConstantTensorDescriptor.hip.hpp"
-#include "ConstantMergedTensorDescriptor.hip.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "ConstantMergedTensorDescriptor.hpp"
 
 template <class Float,
           class SrcDesc,

src/include/threadwise_tensor_slice_op.hip.hpp → src/include/threadwise_tensor_slice_op.hpp

 #pragma once
-#include "ConstantTensorDescriptor.hip.hpp"
+#include "ConstantTensorDescriptor.hpp"
 
 // need to assume src and dst is aligned
 template <class Float, class SrcDesc, class DstDesc, class SrcOpLengths, index_t DataPerRead>

src/include/vector_type.hip.hpp → src/include/vector_type.hpp

 #pragma once
 #include "config.h"
-#include "integral_constant.hip.hpp"
+#include "integral_constant.hpp"
 
 template <class T, index_t N>
 struct vector_type