Andreas Klöckner

That sounds reasonable. I'd be happy to take a PR.

I think what you're encountering is that `buffer_array` (along with `precompute`) currently lay out the buffer in the same way as the underlying array. Hermite normal form is, AFAIK, the...

CI says the failure is in the test you added: ``` _ test_array_method[

I should ask, could you explain the specific need filled by having `asarray` work? Is it just convenience or something deeper?

* Your code is still failing the Flake8 linter, see CI. FWIW, it is expected that your code passes CI before it gets merged. * What will this do for...

Could you please send a message to the mailing list about this? It's likely that you'll get better feedback there. One possible issue is that this uses the runtime (not...

With bd1ea33 (#422) and `lpbug.py` the updated reproducer above: ``` PYTHONHASHSEED=22 pycl lpbug.py ``` I get ``` ------------------------------------------------------------------------------- CODE: ------------------------------------------------------------------------------- __kernel void __attribute__ ((reqd_work_group_size(1, 1, 1))) loopy_kernel(__global int *__restrict__ a,...

I guess I agree with you. Given that this code is getting replaced, let's not worry about it.

As far as I remember, Loopy's system solving does not attempt divisions, so I'm not surprised this doesn't work. That functionality should really just use isl.

The only platform with multiple devices I have access to is a Linux CUDA deployment, everything works there (as in your 2-GPU case). Also, since PyOpenCL doesn't distinguish between GPU...