Deprecation? about compyte

[inducer]

It's not really deprecated--it's an implementation detail of pycuda and pyopencl. Arnaud and Frederic have just decided to take their array and make it a separate thing.

[cancan101]

Long term is the plan for pyopencl and pycuda to use libgpuarray?

[inducer]

The plan for PyOpenCL is for the arrays and the algorithms to be split from the core wrapper. Then libgpuarray would be a competing array package, hopefully with a compatible interface.

I'm less certain what to do on the PyCUDA end, but I'm open to suggestions/patches.

[cancan101]

How is pyopencl different from pycuda in this regard?
On Jul 12, 2014 3:50 PM, "Andreas Klöckner" [email protected]
wrote:

The plan for PyOpenCL is for the arrays and the algorithms to be split
from the core wrapper. Then libgpuarray would be a competing array package,
hopefully with a compatible interface.

I'm less certain what to do on the PyCUDA end, but I'm open to
suggestions/patches.

—
Reply to this email directly or view it on GitHub
#22 (comment).

[inducer]

They're just separate code bases.

[cancan101]

I just meant why is it less certain what to do with pycuda?
On Jul 12, 2014 4:01 PM, "Andreas Klöckner" [email protected]
wrote:

They're just separate code bases.

—
Reply to this email directly or view it on GitHub
#22 (comment).

[inducer]

Somebody would have to do it, as I don't foresee having the time.

[cancan101]

Understood. From a techlogical perspective, however, there should be no reason why the array component should be separable in the case of pyopencl but not pycuda?

[inducer]

No, sure. There's no technical reason here-it's virtually the same code after all.