darshkaushik / HiPC-Programming-Contest-2021

https://hipc.org/programming/

HiPC Programming Contest 2021

Track B: Nvidia GPU Architecture

---

Counting k-cliques in a graph using GPUs.

---

Compilation and Execution Instructions

Required gcc version: 8.1 or above
Required nvcc version: 11.2 or 9.2

Follow these steps to run the solution on a GPU device:

1. Clone or download this Repository.

2. The final solutions are present in the Final Source Codes folder. It has three solutions: optimised CPU solution, GPU solution 1 and GPU solution 2.

3. To run the solutions, create a repository named test (it can be named something else as well). Change the working directory to test.

4. Create a solution.cu file and a input.txt file. Copy the solution to be executed in the solution.cu and the input in the input.txt. Make sure that the input format is correct.

5. Execute the solution.cu file using the command below. The output will be printed on the command-line screen.

$nvcc solution.cu 
$./a.out

If this doesn't work use:

$nvcc -arch=sm_35 -rdc=true solution.cu
$./a.out

Here arch=sm_35 denotes the compute capability of the gpu. For compute capability of 6.0 use sm_60, similarly for 8.5 use sm_85.

---

Input Format

The first line of input should contain two integers - the first one for the number of edges in the graph and the second one for the value of k for which the number of k-cliques needs to be found.

Next lines contain two integers each which represents an edge.

---

Output Format

The number of k-cliques found along with the execution time is printed.

---