Calling C++ CUDA code from python

Question

I want to call a function written in CUDA(C++) from python and pass to it numpy arrays as input and get output arrays from this function. Is this possible? This is the sole objective of this question.

ps : As I understand if such a function were to be made into a proper program and compiled into an executable then it could be called from python as a shell command. However how could I pass and get arrays as inputs and outputs without using files?

Answer 1

PyCuda is provided by nvidia. it's also very easy to use. check it out https://developer.nvidia.com/pycuda

I suppose you want something like that:

import pycuda.autoinit
import pycuda.driver as drv
import numpy

from pycuda.compiler import SourceModule
mod = SourceModule("""
__global__ void multiply_them(float *dest, float *a, float *b)
{
  const int i = threadIdx.x;
  dest[i] = a[i] * b[i];
}
""")

multiply_them = mod.get_function("multiply_them")

a = numpy.random.randn(400).astype(numpy.float32)
b = numpy.random.randn(400).astype(numpy.float32)

dest = numpy.zeros_like(a)
multiply_them(
        drv.Out(dest), drv.In(a), drv.In(b),
        block=(400,1,1), grid=(1,1))

print dest-a*b

Answer 2

有一些库可能对这类任务有所帮助，但PyCuda似乎与你所描述的最接近。