Update CUTLASS and align build flags with newer headers

CMakeLists.txt

@@ -73,9 +73,10 @@ endif()
 if (MSVC)
 	set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /D_CRT_SECURE_NO_WARNINGS")
 	set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /MP")
+	set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zc:preprocessor")
 endif()
 
-set(CMAKE_CXX_STANDARD 14)
+set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_CXX_EXTENSIONS OFF)
 
@@ -134,7 +135,7 @@ function(TCNN_AUTODETECT_CUDA_ARCHITECTURES OUT_VARIABLE)
 	endif()
 endfunction()
 
-set(CMAKE_CUDA_STANDARD 14)
+set(CMAKE_CUDA_STANDARD 17)
 set(CMAKE_CUDA_STANDARD_REQUIRED ON)
 set(CMAKE_CUDA_EXTENSIONS OFF)
 set(CUDA_LINK_LIBRARIES_KEYWORD PUBLIC)
@@ -278,6 +279,7 @@ endif()
 
 if (MSVC)
 	list(APPEND CUDA_NVCC_FLAGS "-Xcompiler=/bigobj")
+	list(APPEND CUDA_NVCC_FLAGS "-Xcompiler=/Zc:preprocessor")
 else()
 	list(APPEND CUDA_NVCC_FLAGS "-Xcompiler=-Wno-float-conversion")
 	list(APPEND CUDA_NVCC_FLAGS "-Xcompiler=-fno-strict-aliasing")

README.md

@@ -161,7 +161,7 @@ producing an image every couple of training steps. Each 1000 steps should take a
 ## Requirements
 
 - An __NVIDIA GPU__; tensor cores increase performance when available. All shown results come from an RTX 3090.
-- A __C++14__ capable compiler. The following choices are recommended and have been tested:
+- A __C++17__ capable compiler. The following choices are recommended and have been tested:
   - __Windows:__ Visual Studio 2019 or 2022
   - __Linux:__ GCC/G++ 8 or higher
 - A recent version of __[CUDA](https://developer.nvidia.com/cuda-toolkit)__. The following choices are recommended and have been tested: