reorginzed files

include/composable_kernel/utility/functional2.hpp → composable_kernel/include/utility/functional2.hpp

 #ifndef CK_FUNCTIONAL2_HPP
 #define CK_FUNCTIONAL2_HPP
 
-#include "composable_kernel/utility/functional.hpp"
-#include "composable_kernel/utility/Sequence.hpp"
+#include "functional.hpp"
+#include "Sequence.hpp"
 
 namespace ck {
 

include/composable_kernel/utility/functional3.hpp → composable_kernel/include/utility/functional3.hpp

 #ifndef CK_FUNCTIONAL3_HPP
 #define CK_FUNCTIONAL3_HPP
 
-#include "composable_kernel/utility/functional.hpp"
-#include "composable_kernel/utility/functional2.hpp"
-#include "composable_kernel/utility/Sequence.hpp"
-#include "composable_kernel/utility/Array.hpp"
+#include "functional.hpp"
+#include "functional2.hpp"
+#include "Sequence.hpp"
+#include "Array.hpp"
 
 namespace ck {
 

include/composable_kernel/utility/utility.hpp → composable_kernel/include/utility/utility.hpp

 #ifndef CK_UTILITY_HPP
 #define CK_UTILITY_HPP
 
+#include "config.hpp"
+
 namespace ck {
 
 __device__ index_t get_thread_local_1d_id() { return threadIdx.x; }

include/composable_kernel/utility/vector_type.hpp → composable_kernel/include/utility/vector_type.hpp

 #ifndef CK_VECTOR_TYPE_HPP
 #define CK_VECTOR_TYPE_HPP
 
-#include "composable_kernel/utility/config.hpp"
-#include "composable_kernel/utility/integral_constant.hpp"
+#include "config.hpp"
+#include "integral_constant.hpp"
 
 namespace ck {
 

driver/CMakeLists.txt

+set(TENSOR_SOURCE 
+    src/tensor.cpp;
+    src/device.cpp;
+)
+
+add_library(tensor SHARED ${TENSOR_SOURCE})
+target_compile_features(tensor PUBLIC)
+set_target_properties(tensor PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
+if(DEVICE_BACKEND STREQUAL "NVIDIA")
+    target_link_libraries(tensor nvToolsExt cudart)
+endif()
+
+install(TARGETS tensor LIBRARY DESTINATION lib) 
+
+
 if(DEVICE_BACKEND STREQUAL "AMD")
-    set(DRIVER_SOURCE driver.cpp)
+    set(DRIVER_SOURCE src/driver.cpp)
 elseif(DEVICE_BACKEND STREQUAL "NVIDIA")
-    set(DRIVER_SOURCE driver.cu)
+    set(DRIVER_SOURCE src/driver.cu)
 endif()
 
 add_executable(driver ${DRIVER_SOURCE}) 

include/conv_common.hpp → driver/include/conv_common.hpp

 #ifndef CK_CONV_COMMON_HPP
 #define CK_CONV_COMMON_HPP
 
-#include "composable_kernel/tensor_description/ConstantTensorDescriptor.hpp"
+#include "ConstantTensorDescriptor.hpp"
 
 using namespace ck;
 

include/device.hpp → driver/include/device.hpp

@@ -2,7 +2,7 @@
 #define CK_DEVICE_HPP
 
 #include <memory>
-#include "composable_kernel/utility/config.hpp"
+#include "config.hpp"
 
 using namespace ck;
 

driver/device_convolution_direct_v2_nchw_kcyx_nkhw.hpp → driver/include/device_convolution_direct_v2_nchw_kcyx_nkhw.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
+#include "tensor.hpp"
 #include "gridwise_convolution_kernel_wrapper.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_direct_v2_nchw_kcyx_nkhw.hpp"
+#include "gridwise_convolution_direct_v2_nchw_kcyx_nkhw.hpp"
 
 using namespace ck;
 

driver/device_convolution_implicit_gemm_v1_chwn_cyxk_khwn.hpp → driver/include/device_convolution_implicit_gemm_v1_chwn_cyxk_khwn.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
+#include "tensor.hpp"
 #include "gridwise_convolution_kernel_wrapper.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v1r1_chwn_cyxk_khwn.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v1r2_chwn_cyxk_khwn.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v1r3_chwn_cyxk_khwn.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v1r3_chwn_cyxk_khwn_lds_double_buffer.hpp"
+#include "gridwise_convolution_implicit_gemm_v1r1_chwn_cyxk_khwn.hpp"
+#include "gridwise_convolution_implicit_gemm_v1r2_chwn_cyxk_khwn.hpp"
+#include "gridwise_convolution_implicit_gemm_v1r3_chwn_cyxk_khwn.hpp"
+#include "gridwise_convolution_implicit_gemm_v1r3_chwn_cyxk_khwn_lds_double_buffer.hpp"
 
 using namespace ck;
 

driver/device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw.hpp → driver/include/device_convolution_implicit_gemm_v1_nchw_cyxk_nkhw.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
+#include "tensor.hpp"
 #include "gridwise_convolution_kernel_wrapper.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v1r3_nchw_cyxk_nkhw.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v1r3_nchw_cyxk_nkhw_lds_double_buffer.hpp"
+#include "gridwise_convolution_implicit_gemm_v1r3_nchw_cyxk_nkhw.hpp"
+#include "gridwise_convolution_implicit_gemm_v1r3_nchw_cyxk_nkhw_lds_double_buffer.hpp"
 
 using namespace ck;
 

driver/device_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hpp → driver/include/device_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
+#include "tensor.hpp"
 #include "gridwise_convolution_kernel_wrapper.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn_lds_double_buffer.hpp"
+#include "gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hpp"
+#include "gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn_lds_double_buffer.hpp"
 
 using namespace ck;
 

driver/device_convolution_implicit_gemm_v3_nchw_cyxk_nkhw.hpp → driver/include/device_convolution_implicit_gemm_v3_nchw_cyxk_nkhw.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
+#include "tensor.hpp"
 #include "gridwise_convolution_kernel_wrapper.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v3_nchw_cyxk_nkhw.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v3_nchw_cyxk_nkhw_lds_double_buffer.hpp"
+#include "gridwise_convolution_implicit_gemm_v3_nchw_cyxk_nkhw.hpp"
+#include "gridwise_convolution_implicit_gemm_v3_nchw_cyxk_nkhw_lds_double_buffer.hpp"
 
 using namespace ck;
 

driver/device_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hpp → driver/include/device_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
+#include "tensor.hpp"
 #include "gridwise_convolution_kernel_wrapper.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_convolution_implicit_gemm_v4_nchw_kcyx_nkhw_lds_double_buffer.hpp"
+#include "gridwise_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hpp"
+#include "gridwise_convolution_implicit_gemm_v4_nchw_kcyx_nkhw_lds_double_buffer.hpp"
 
 using namespace ck;
 

driver/device_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hpp → driver/include/device_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hpp"
+#include "tensor.hpp"
+#include "gridwise_direct_convolution_2_vectorized_nchw_kcyx_nkhw.hpp"
 
 using namespace ck;
 

driver/device_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hpp → driver/include/device_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hpp

 #pragma once
 #include <unistd.h>
 #include "device.hpp"
-#include "composable_kernel/kernel_algorithm/gridwise_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hpp"
+#include "tensor.hpp"
+#include "gridwise_implicit_gemm_convolution_1_chwn_cyxk_khwn_padded.hpp"
 
 using namespace ck;
 

src/device.cpp → driver/src/device.cpp

-#include "composable_kernel/utility/config.hpp"
+#include "config.hpp"
 #include "device.hpp"
 
 DeviceMem::DeviceMem(std::size_t mem_size) : mMemSize(mem_size)

driver/driver.cpp → driver/src/driver.cpp

@@ -3,9 +3,9 @@
 #include <initializer_list>
 #include <cstdlib>
 #include <stdlib.h>
-#include "composable_kernel/utility/config.hpp"
-#include "composable_kernel/tensor_description/ConstantTensorDescriptor.hpp"
-#include "tensor.hpp"
+#include "config.hpp"
+#include "ConstantTensorDescriptor.hpp"
+#include "device.hpp"
 #include "conv_common.hpp"
 #include "device_convolution_direct_v2_nchw_kcyx_nkhw.hpp"
 #include "device_convolution_implicit_gemm_v1_chwn_cyxk_khwn.hpp"