Luca Salucci <lucasalu...@libero.it> wrote on 02/18/2014 03:23:55 AM:
>
> Hello, I was wondering if it is possible to move a type Rail[Rail
> [Int]] from host to GPU and to handle it within the kernel, both in
> shared or constant memory (I need only to read it). I tried in
> different ways but I still do not manage to compile.
>
Hi,
Sorry for the slow response. I believe what is currently supported
is really just primitives (int, float, long, etc) and Rails of the same
primitives. Rails of Rails are not supported.
It is not currently supported, but I don't see too many obstacles to
supporting x10.array.Array_2 or Array_3 in an @CUDA kernel. The backing
store for Array_2 is just a Rail with a straightforward mapping of (i,j)
array indices into an offset. Would this handle your usage, or do you
really need the flexibility of Rail[Rail[T]] ?
--dave
------------------------------------------------------------------------------
Managing the Performance of Cloud-Based Applications
Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
Read the Whitepaper.
http://pubads.g.doubleclick.net/gampad/clk?id=121054471&iu=/4140/ostg.clktrk
_______________________________________________
X10-users mailing list
X10-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/x10-users
