Creating a CUDA function with PyCUDA (pt. 1)

pycuda_exampple_pt1.py

import numpy as np
import pycuda.driver as cuda
import pycuda.autoinit
from pycuda.compiler import SourceModule

# Define our function using CUDA code
cuda_func_def = """
  __global__ void multiply(float *result, float *a, float *b)
  {
    const int i = threadIdx.x;
    result[i] = a[i] * b[i];
  }
"""
        
# Create CUDA module and import into Python
mod = SourceModule(cuda_func_def)
multiply_func = mod.get_function("multiply")