numbapro中的cuda代码错误

Question

import numpy
import numpy as np
from numbapro import cuda


@cuda.autojit
def foo(aryA, aryB,out):
    d_ary1 = cuda.to_device(aryA)
    d_ary2 = cuda.to_device(aryB)
    #dd = numpy.empty(10, dtype=np.int32)
    d_ary1.copy_to_host(out)


griddim = 1, 2
blockdim = 3, 4
aryA = numpy.arange(10, dtype=np.int32)
aryB = numpy.arange(10, dtype=np.int32)
out = numpy.empty(10, dtype=np.int32)

foo[griddim, blockdim](aryA, aryB,out)

Exception: Caused by input line 11: can only get attribute from globals, complex numbers or arrays

I am new to numbapro, hints are needed!

Answer 1

The @cuda.autotjit marks and compiles foo() as a CUDA kernel. The memory transfer operations should be placed outside of the kernel. It should look like the following code:

import numpy
from numbapro import cuda

@cuda.autojit
def foo(aryA, aryB ,out):
    # do something here
    i = cuda.threadIdx.x + cuda.blockIdx.x * cuda.blockDim.x
    out[i] = aryA[i] + aryB[i]

griddim = 1, 2
blockdim = 3, 4
aryA = numpy.arange(10, dtype=numpy.int32)
aryB = numpy.arange(10, dtype=numpy.int32)
out = numpy.empty(10, dtype=numpy.int32)

# transfer memory
d_ary1 = cuda.to_device(aryA)
d_ary2 = cuda.to_device(aryB)
d_out = cuda.device_array_like(aryA) # like numpy.empty_like() but for GPU
# launch kernel
foo[griddim, blockdim](aryA, aryB, d_out)

# transfer memory device to host
d_out.copy_to_host(out)

print out

I recommend new NumbaPro users to look at the examples in https://github.com/ContinuumIO/numbapro-examples .