On Fri, 28 Oct 2011 11:48:57 +0200, Peter17 <peter017(a)gmail.com> wrote:
> Date: Sun, 14 Aug 2011 15:30:56 +0200
> From: Andreas Kloeckner <lists(a)informa.tiker.net>
> sorry for the long delay in responding. For now, pass the 'cuda_no_plan'
> debug flag, i.e. add (..., debug=["cuda_no_plan"]) to
> make_discretization. The CUDA and CPU ('jit') backends will disappear in
> relatively short order (one or two months), giving way to a loopy
> -assisted OpenCL backend that will cover both CPU and GPU uses. I'm
> not sure it makes putting much work into the CUDA code for that reason.
This sounds very interesting to me! I would be glad to test it and
give help if I can.
Is there already a way to use loopy as a backend for Hedge? Can you
please indicate me a procedure to do so?
The short version is that loopy is only just coming online, and there's
no code in hedge yet to target it. The slightly longer version version
is that loopy was already almost done in early September, but I decided
to redo it from scratch to make it more useful on more (and different)
problems. This work is only just bearing fruit. I'll let you guys know
once I have a branch of hedge that's worth testing.