Greetings!
We’re about to release a graph library for GPUs. The library compiles fine on,
well, everything except clang (I’m using OS X), and it’s because clang does not
support the same template feature set as other compilers. ("error: non-type
template argument depends on a template parameter of the partial
specialization")
Here’s the issue and the reply from Duane Merrill at NVIDIA:
https://github.com/gunrock/gunrock/issues/1
(I’ll append a short version of it below.)
What I’d love to know is “is this something that clang will support in the near
future / is there an easy clang workaround”. If neither is true, perhaps we do
some refactoring, per Duane’s suggestion. But if either is true, that would
ease our support issues.
JDO
load_tile.cuh:207:
template <int LOAD, int dummy>
struct Iterate<LOAD, LOAD_VEC_SIZE, dummy>
The error I'm getting is:
../../gunrock/util/io/load_tile.cuh:207:23: error: non-type template argument
depends on a template parameter of the partial specialization
_______________________________________________
cfe-users mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-users
