README.md

Gridder- CUDA Implementation

Note: currently only supported by NVIDIA GPUs (limitation of CUDA)

---

Instructions for installation of this software (includes profiling, linting, building, and unit testing):

1. Ensure you have an NVIDIA based GPU (mandatory!)
2. Install the CUDA toolkit and runtime (refer to link for download/installation procedure)
3. Install Valgrind (profiling, memory checks, memory leaks etc.)

   $ sudo apt install valgrind

4. Install Cmake/Makefile (build tools)

   $ sudo apt install cmake

5. Install Google Test (unit testing) - See this tutorial for tutorial on using Google Test library

   $ sudo apt install libgtest-dev
   $ cd /usr/src/gtest
   $ sudo cmake CMakeLists.txt
   $ sudo make
   $ sudo cp *.a /usr/lib

6. Install Cppcheck (linting)

   $ sudo apt install cppcheck

7. Configure the code for usage (modify gridder config)
8. Create local execution folder

   $ mkdir build && cd build

9. Build gridder project (from project folder)

   $ cmake .. -DCMAKE_BUILD_TYPE=Release && make

10. Important: set -CDMAKE_BUILD_TYPE=Debug if planning to run Valgrind. Debug mode disables compiler optimizations, which is required for Valgrind to perform an optimal analysis.

---

Instructions for usage of this software (includes executing, testing, linting, and profiling):

To perform memory checking, memory leak analysis, and profiling using Valgrind, execute the following (assumes you are in the appropriate build folder (see step 5 above):

$ valgrind --leak-check=yes -v ./gridder

To execute linting, execute the following commands (assumes you are in the appropriate source code folder):

$ cppcheck --enable=all main.cpp
$ cppcheck --enable=all gridder.cu

To execute the gridder (once configured and built), execute the following command (also assumes appropriate build folder):

$ ./gridder