Untitled

#!/usr/bin/env python

import pycuda.autoinit
import pycuda.driver as cuda
import numpy

a = numpy.array([ [1., 2., 1., 1.],
                  [1., 1., 1., 1.],
                  [1., 1., 1., 1.],
                  [1., 1., 6., 1.] ]).astype(numpy.float32)

a_gpu = cuda.mem_alloc(a.nbytes)
cuda.memcpy_htod(a_gpu, a)

mod = cuda.SourceModule("""
  __global__ void doublify(float *a)
  {
    int idx = threadIdx.x + threadIdx.y * 4;
    a[idx] *= 2;
  }
  """)

func = mod.get_function("doublify")
func(a_gpu, block=(4,4,1))

a_doubled = numpy.empty_like(a)
cuda.memcpy_dtoh(a_doubled, a_gpu)

print "Input:"
print a

print "Output:"
print a_doubled