Hi Andreas,
On 17/02/2012 23:21, Andreas Kloeckner wrote:
Hi Jon,
On Fri, 17 Feb 2012 19:05:30 +0100, Jon Wright<[email protected]> wrote:
The patches I described here for pyopencl also work on pyopencl as
bpl-subset is the same:
http://www.mail-archive.com/[email protected]/msg02285.html
If you could confirm whether these work for you too and look correct,
perhaps we could update and wiki and try to get the official trees fixed.
Sorry, last time around I simply missed that there were patches to merge
there. I've merged your changes to aksetup. Can you explain your tss
cleanup changes? Those didn't necessarily seem harmless, so I would like
to understand what I'm doing before I grab them. Also, I believe I've
updated boost (in git) since your patch--can you please verify that it's
still necessary?
I'm afraid the tss issue is still there on a clean checkout of pycuda or
pyopencl and bpl-subset:
thread.obj : error LNK2019: unresolved external symbol "void __cdecl
pycudaboost::tss_cleanup_implemented(void)"...
thread.obj : error LNK2019: unresolved external symbol "void __cdecl
pyopenclboost::tss_cleanup_implemented(void)"
If I comment out line 33 of bpl-subset/libs/thread/src/win32/thread.cpp
then I am able to compile and link and tests all pass for pycuda with
CUDA 4.1. PyOpenCl runs my own code but has multiple issues in the test
suite that I'll get to in a separate mail.
For the tss I guess we need to ask a boost expert?
Best
Jon
_______________________________________________
PyOpenCL mailing list
[email protected]
http://lists.tiker.net/listinfo/pyopencl
