for validation of results and speed comparisons it's essential to be able to generate RNGs on CPU as well as device, so this should be supported by the library.
Id #1212 | Release:
| Updated: Feb 13, 2013 at 8:57 PM by DJMcCrady | Created: Jun 20, 2012 at 2:42 AM by terryspitz
i think you should add accelerator_view to your base class and a defaulted version to the derived:
const concurrency::extent<_rank> rand_extent,
Id #1211 | Release:
| Updated: Jan 17, 2015 at 12:15 AM by mbenedict115 | Created: Jun 20, 2012 at 2:17 AM by terryspitz