/gpu-ntt

Number Theoretic Transform Implementation on GPU for FHE Applications

Primary LanguageCuda

gpu-ntt

Number Theoretic Transform Implementation on GPU for FHE Applications

File Descriptions:

helper.h

includes:

  • modular power calculation
  • reversing bits
  • random array creation
  • calculating twiddle factors

ntt_30bit.cuh

includes gpu functions:

  • barrett
  • ntt
  • intt

30bit_ntt_test.cu

includes:

the main program for applying ntt, then intt on a randomly generated array

How to run

Compile with: nvcc -arch=sm_XX -rdc=true -cudart static --machine 64 -use_fast_math -O2 30bit_ntt_test.cu -o 30bit_ntt -lcudadevrt -std=c++11

(replace XX with the compute capability of your GPU)

Run with: ./30bit_ntt

Coded by: Can Elgezen, Özgün Özerk
Contributed by: Ahmet Can Mert, Erkay Savaş, Erdinç Öztürk