Nice! Would you contribute your fix?
Peter
Am 10.10.2018 18:33 schrieb Nil Goyette :
Hi Peter,
Surprisingly I fixed it. In 3 easy steps :) I don't even know if all cores are
used, but it does compile and start.
Cherry-pick your 4 commits, fix conflicts
0e19e394243f Omp variable int and omp.h
Hi Peter,
Surprisingly I fixed it. In 3 easy steps :) I don't even know if all
cores are used, but it does compile and start.
*Cherry-pick your 4 commits, fix conflicts*
0e19e394243f Omp variable int and omp.h includes
82a46f59a410 Replace reduction by critical
3fc94be7814c Add openmp depen
This is where I'm stuck. Didn't have the time to investigate further since I'm
out of office for a while.
Peter
Am 09.10.2018 15:42 schrieb Nil Goyette :
Hi Peter,
What do you mean by "unsuccessful until now"? I tested
T25191-v2018.04-beta, cherry-picking your 4 OpenMP commits and it's
still un
Hi Peter,
What do you mean by "unsuccessful until now"? I tested
T25191-v2018.04-beta, cherry-picking your 4 OpenMP commits and it's
still unsuccessful. I have a list of undefined symbols, for example
Undefined symbols for architecture x86_64:
"___kmpc_critical", referenced from:
_.omp
Hi Nil,
The trick is that we do not build MITK Diffusion for Mac, which is admittedly a
rather poor solution. Apple Clang does not ship with openmp anymore. The branch
you mentioned was a first try to fix this following this solution:
https://iscinumpy.gitlab.io/post/omp-on-high-sierra/
Unfortu
Hi all, Peter,
I'm trying to build MITK beta on macOS (all tools installed/upgraded
with mac port) and it's failing in the DiffusionImaging module because
of OpenMP.
.../Modules/DiffusionImaging/FiberTracking/Algorithms/itkStreamlineTrackingFilter.cpp:21:10:
fatal error: 'omp.h' file not fou
