Hi Karl,
Karl Rupp <r...@iue.tuwien.ac.at> writes:
> Sure. Could you please derive some simple test case from
> examples/tutorial/matrix-range.cpp with slices in order to find the
> cause of the issue? You should be able to implement the submatrix proxy
> described above in a straight-forward manner *and* get correct results :-)
I soon suspected that the bug was in the copying of data from the GPU
handle to the ndarray: operations on the proxies were reflected in the
original matrix, and deriving a test case form matrix-range.cpp just
confirmed that. So I'm working on fixing my copying to do more than just
vcl::backend::memory_read(m.handle(), 0,
(rows*cols*sizeof(SCALARTYPE)),
data);
I'm going to base my code on the copy function in matrix_proxy.hpp -- I
can't use that directly because I'm only exposing the proxy types to
Python as *_base instances (I'm trying to keep my EXPOSE_* macros as
manageable as possible, which means exposing as few derived C++ types as
I can get away with).
Thanks for the help!
Toby
------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead.
Download for free and get started troubleshooting in minutes.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
ViennaCL-devel mailing list
ViennaCL-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/viennacl-devel
