deploy: b1f5bd3

Author: jardiamj

.nojekyll

@@ -0,0 +1,27 @@
+# Definitions
+
+- Top level configuration file: CMakeLists.txt
+- Enable_language: enables stuff built-in to CMake
+- CMake documentation at (example for HDF5): https://cmake.org/cmake/help/v3.12/module/FindHDF5.html
+- Add_compile_definitions: preprocessor define
+- Include_directories: just for .h
+- Add_subdirectory: contain other CMakeLists.txt
+- Add_executable: what to generate, the main() .cpp
+- Target_link_libraries: indicates what needs to be linked with a target (1st arg).
+- Add_library: creating symbol (name for a bunch of code)
+- File: generates a list of files, 1st arg is name of list
+- List: manipulation utility
+- Find_package: loads a package that comes with CMake (like HDF5)
+
+
+# Current Configuration
+
+
+# Avenues for Future Enhancement
+
+
+# Hints
+
+- Verbose output: https://sidvind.com/wiki/CMake/Verbose_output
+
+

docs/Developer/CMake.md

@@ -0,0 +1,42 @@
+
+# Using CUDA and CMakeLists  
+
+This section contains information on a few things found to be necessary for the  
+simulator CUDA code to build and run correctly when using CMakeLists. This is  
+not a complete guide on using CMakeLists with CUDA.    
+
+## Static Libraries  
+
+Each library created in the CMakeLists.txt file that contains CUDA source code  
+should be created with the STATIC keyword. More information on the difference  
+between STATIC and SHARED type libraries can be found in the CMake documentation,  
+and a short summary is available on the [add_library documentation page](https://cmake.org/cmake/help/latest/command/add_library.html). Libraries  
+with no type specified will be assigned one automatically based on conditions  
+listed in the documentation, but in the case of our simulator they can be observed  
+as being SHARED if no type is specified.   
+
+If a library containing CUDA source files is created as SHARED, the linking of  
+the corresponding CUDA device code will be done individually just before the  
+step of linking the SHARED library it is a part of. Explicitly declaring libraries  
+containing CUDA code as STATIC instead allows the linking of CUDA device code to  
+be put off and done all together just before the final executable is linked and  
+built. This change allows the linker for CUDA code to function correctly and  
+prevents many errors that arise if the libraries are left as SHARED.  
+
+Example of creating a STATIC library:  
+    ```add_library(Edges STATIC ${Edges_Source})```  
+    Where `${Edges_Source}` is a collection of source files
+ 
+## Separable Compilation  
+
+CUDA separable compilation allows different pieces of CUDA code to be compiled  
+into separate objects and then linked together later. More information can be  
+found about Separate Compilation and Linking of CUDA code in this [NVIDIA developer  
+blog post](https://developer.nvidia.com/blog/separate-compilation-linking-cuda-device-code/). This must be enabled in the CMakeLists.txt file individually for each  
+library containing CUDA source code by setting the CUDA_SEPARABLE_COMPILATION  
+property to ON. Information on this property can be found [here](https://cmake.org/cmake/help/latest/prop_tgt/CUDA_SEPARABLE_COMPILATION.html) in the CMake  
+documentation.  
+
+It is set using the set_property command, for example like this:  
+    ```set_property(TARGET Edges PROPERTY CUDA_SEPARABLE_COMPILATION ON)```  
+    Where `Edges` is a library containing CUDA source code.