Using openMP in the cuda host code?

Question

It it possible to use openMP pragmas in the CUDA-Files (not in the kernel code)?

I will combine gpu and cpu computation. But nvvc compiler fails with "cannot find Unknown option 'openmp' ", if i am linking the porgram with a openmp option (under linux)

A wayaround is to use openMP-statments only in c/c++ files.

score 21 · Accepted Answer · edited May 23 '14 at 19:43

21

I've just found this

http://www.cse.buffalo.edu/faculty/miller/Courses/CSE710/heavner.pdf

Page 25 says:

With gcc: -#include omp.h

Add the -fopenmp flag

With nvcc, this should be -Xcompiler -fopenmp as this needs to be passed directly to gcc -Xcompiler passes flags directly to host compiler

Add -lgomp flag during the linking stage.

I haven't tried it yet...

edited May 23 '14 at 19:43

Doug Lipinski

644
4
9

answered Jul 19 '10 at 15:51

Javier Ruiz

410
4
10

Be careful coping and pasting the arguments. If you look closely the dashes on -Xcompiler -fopenmp are different from command line dashes. – Tristan Oct 04 '13 at 22:07
The dashes are fixed now. – Doug Lipinski May 23 '14 at 19:44

score 2 · Answer 2 · answered Oct 20 '12 at 20:09

I tried writing the parameter in "Additional Compiler Options" but it didn't work.

What I did for Visual Studio 2010 and CUDA 4.2:

In Project Properties -> Configuration Properties -> CUDA C/C++ -> Command Line -> Additional Options: -Xcompiler "/openmp"

This resulted in two -Xcompiler parameters in the resulting build command but did not cause any problems and worked successfully.

score 0 · Answer 3 · answered Jul 22 '10 at 09:58

0

The solution for the Visual Studio found from the nvidia-forum:

add the '/openmp'-flag to the Extra C++ Options in cuda build rules. I will try the linux solution later.

answered Jul 22 '10 at 09:58

LonliLokli

1,295
4
15
24

score 0 · Answer 4 · answered Nov 10 '22 at 11:27

Building with CMake

I had to add -Xcompiler=-fopenmp as an compile option to my CMakeLists.txt file to build CUDA host code with OpenMP directives:

# your CMakeLists.txt should contain something like this already
project(<project> Languages CXX CUDA)
find_package(CUDA REQUIRED)
find_package(OpenMP)
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${OpenMP_C_FLAGS}")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS}")

# the following line was also necessary
target_compile_options(<target> PRIVATE $<$<COMPILE_LANGUAGE:CUDA>: -Xcompiler=-fopenmp>)

Using openMP in the cuda host code?

4 Answers4

Building with CMake

Linked