Giter VIP home page Giter VIP logo

Comments (6)

sjeaugey avatar sjeaugey commented on August 20, 2024

The $MPI_HOME variable in the Makefileassumes includes are in $MPI_HOME/include and libraries are in $MPI_HOME/lib. In most cases, /usr/include/mpi is just a link, and the real MPI installation is in /usr/lib/openmpi (both includes and library). Setting MPI_HOME=/usr/lib/openmpi would solve the problem.

Another simple solution is to add /usr/include/mpi to the path of g++ includes :

export CPLUS_INCLUDE_PATH=/usr/include/mpi:$CPLUS_INCLUDE_PATH

And if the libraries are not found (cannot find -lmpi), we can also tell g++ where to find libraries :

export LIBRARY_PATH=/usr/lib/openmpi/lib:$LIBRARY_PATH

(but usually there is a link in /usr/lib/libmpi.so so it shouldn't be needed)

from nccl-tests.

scottzockoll avatar scottzockoll commented on August 20, 2024

Hello,
I am using a Jetson Nano cluster and /usr/lib/openmpi does not exist. It seems the mpi is installed because I get the following output when I run mpirun --version:

mpirun (Open MPI) 2.2.1

Report bugs to http://www.open-mpi.org/community/help/

from nccl-tests.

sjeaugey avatar sjeaugey commented on August 20, 2024

On ARM platforms, the path to Open MPI might be something like /usr/lib/aarch64-linux-gnu/openmpi.

mpicc -show would show where MPI includes and library are located, then you can use MPI_HOME or CPLUS_INCLUDE_PATH/LIBRARY_PATH to make sure g++ finds them well.

Please note however that NCCL is not targeted at Jetson Nano so I would not expect it to work.

from nccl-tests.

scottzockoll avatar scottzockoll commented on August 20, 2024

I was able to find it under /usr/lib/aarch64-linux-gnu/openmpi However when running the command mpirun -np 16 ./build/all_reduce_perf -b 32M -e 128M -f 2 -t 1 -g 1 -c 0 the following error ouputs:

./build/all_reduce_perf: error while loading shared libraries: libnccl.so.2: cannot open shared object file: No such file or directory

But from my understanding this should be found under the NCCL_HOME variable?

from nccl-tests.

sjeaugey avatar sjeaugey commented on August 20, 2024

If you did not install NCCL on the system (just compiled it in your home directory), you need to set LD_LIBRARY_PATH :

export LD_LIBRARY_PATH=$NCCL_HOME/build/lib:$LD_LIBRARY_PATH

from nccl-tests.

AddyLaddy avatar AddyLaddy commented on August 20, 2024

Closing old issues. Please open another one if you are still seeing issues.

from nccl-tests.

Related Issues (20)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.