[FEA] Supporting cuDF Series in device_array_like

[jakirkham]

Is your feature request related to a problem? Please describe.

I'd like to run the following code.

from librmm_cffi import librmm as rmm
import cudf

s = cudf.Series([0, 1, 2])
a = rmm.device_array_like(s)

Currently this fails with the following error.

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-3-917a74c3463e> in <module>
----> 1 rmm.device_array_like(s)

~/miniconda/envs/rapids9/lib/python3.7/site-packages/librmm_cffi/wrapper.py in device_array_like(self, ary, stream)
    227             ary = ary.reshape(1)
    228 
--> 229         return self.device_array(ary.shape, ary.dtype, ary.strides,
    230                                  stream=stream)
    231 

AttributeError: 'Series' object has no attribute 'strides'

Describe the solution you'd like

It would be great if rmm.device_array_like worked with Series objects. No strong feelings about how that is accomplished.

Describe alternatives you've considered

We could special case handling of Series objects, but this shifts the burden to other libraries to solve this problem.

Alternatively cuDF Series objects could gain a strides attribute. This could be reasonable.

Additional context

This came up when trying to better handle GPU array-like objects in cuML ( rapidsai/cuml#1086 ), which is part of the Grid Search effort.

Edit: More specifically, we tried to use librmm_cffi.librmm.to_device instead of numba.cuda.to_device, but were unable to as Series are not supported.

[jakirkham]

Hmm...the doc label here is a bit unexpected. Unfortunately I don't seem able to fix that. Should be flagged as an enhancement or possibly a bug depending.

[jakirkham]

FWIW if I perform this operation with Numba's to_device, it works without issues.

import numba
import numba.cuda

numba.cuda.to_device(s)

cc @kkraus14 ( in response to this comment rapidsai/cuml#1086 (comment) )

[kkraus14]

I'm guessing numba is using __cuda_array_interface__ here, whereas RMM doesn't since it assumes it should allocate memory.

EDIT: yup

In [28]: cuda.to_device(s).device_ctypes_pointer.value
Out[28]: 139706211778048

In [29]: s.data.mem.device_ctypes_pointer.value
Out[29]: 139706211778048

[jakirkham]

Yeah that makes sense. Using __cuda_array_interface__ here would be great (assuming others agree).

[kkraus14]

Removing array functionality in RMM in future so closing this issue.

[jakirkham]

For future readers, please see issue ( #180 ) for context.