guru-irl / parallel-aes Goto Github PK

Algorithm 1 in paper. Uses CUDA
Take heavy inspiration from the core parallel AES module.
Implement the following modules

• Coalescing
• Slicing

Documentation needs to be added for every algorithm. @gurupunskill will do a sample one for one algorithm. The others must follow the same format and structure.

• Sequential
• GNC
• GCNS
• GCS
• CNC
• CCNS
• CCS
• generate

Furthermore, the main readmes for ever folder should be completed

• Main
• src
• Analysis
• docs

The OpenMP variant of the CUDA Parallel Core

GCS without Slicing. Should be simple once the GCS code is complete.

• Complete Encryption code
• Use files to handle inputs

Use CUDA to parallelize the dataset generation.

This is basically our core parallel AES module.

We need to implement a purely parallel AES algorithm. AES works in 16byte chunks of data. So given an input size of say N kilobytes, we need to split the N kilobytes between multiple threads and execute AES Encryption on each thread.
This is the CUDA core kernel that needs to be called in the 3 other implementations.

• Generate Input files
• Generate Keys per file

OpenMP variant of GCS.
Should Ideally look exactly the same as GCS except for the CUDA bits

• Add time analysis code to every algorithm
  • Sequential
  • GCS
  • CCS
  • GCNS
  • CNCS
  • GNC
  • CNC
• Dump timing data into a csv file
• Use python to visualzse data