I added the following lines to my 'Makefile.am' in the directory with the
CUDA sources
.cu.lo:
/usr/local/cuda/bin/nvcc -gencode arch=compute_60,code=sm_60 -O3
--cuda -c $<
mv -f $*.cu.cpp.ii $*.ii
libtool --mode=compile $(CXX) $(CXXFLAGS) -c $*.ii
and added the CUDA
Hi Chris,
First, you will need to have some configure stuff to detect nvcc and use
it inside your Makefile. UTK may have some examples to show here.
For the C/C++ API, you need to add 'extern "C"' statements around the
interfaces you want to export in C so that you can use them inside Open
It looks like the mailing system deleted the attachment, so here it is
inline
#
# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
# University Research and Technology
# Corporation. All rights reserved.
# Copyright (c)
Here is the complete Makefile so far. I have it in directory
ompi/mca/coll/ibm , which contains an implementation of an IBM-written
collectives library.
It won't work as-is, because I don't know how to use 'libtool' which is
presumably needed to do the compile. If anybody can show me a rule
It's hard to tell without complete makefile example. Could you please post
a minimal reprocase?
Note specifically for OpenMPI there is a tricky workaround. You can use
nvcc as mpicc compiler by exporting OMPI_CC=nvcc and wrapping out
incompatible compiler options.
Kind regards,
- Dmitry
I'm trying to build a CUDA kernel into OpenMPI (because I'm experimenting
with an Allreduce collective with data in GPU buffers, and I want the GPU
to do the reduction).
This involves writing a '.cu' file, and compiling this to '.o' with the
NVIDIA CUDA compiler 'nvcc'; and also writing some
