Day: July 9, 2023

by

Compiling OpenMPI-4.1.5 for ROCEv2 with GNU-8.5

https://docs.open-mpi.org/en/v5.0.x/release-notes/networks.html

Prerequisites 1

First thing first, You may want to check whether you are using RoCE. Do take a look at Installing RoCE using Mellanox (Nvidia) OFED package

Prerequisites 2

Do check whether you have ucx. You can do a dnf install 

# dnf install ucx ucx-devel

Alternatively, you can do a manual install. For information on how to install, do take a look at http://openucx.org/wp-content/uploads/UCX_install_guide.pdf 

# wget https://github.com/openucx/ucx/releases/download/v1.4.0/ucx-1.4.0.tar.gz
$ tar xzf ucx-1.4.0.tar.gz
$ cd ucx-1.4.0
$ ./contrib/configure-release --prefix=/usr/local/ucx-1.4.0
$ make -j8 
$ make install

Prerequisites 3

Make sure you have install GNU and GNU-C++. This can be done easily using the 

# dnf install gcc-c++ gcc

Step 1: Download the OpenMPI package

You can go to OpenMPI to download the latest package at (https://www.open-mpi.org/software/ompi/v4.1/). The latest one at the point of writing is OpenMPI-4.1.

Step 2: Compile the Package

$ ./configure --prefix=/usr/local/openmpi-4.1.5 --enable-mpi-cxx --with-devel-headers --with-ucx --with-verbs --with-slurm=no
$ make && make install

Step 3: To run the MPIRUN using ROCE, do the following.

You may want to see Network Support Information on OpenMPI

$ mpirun --np 12 --hostfile path/to/hostfile --mca pml ucx -x -x UCX_NET_DEVICES=mlx5_0:1 ........

References:

  1. Setting up a RoCE cluster
  2. OpenMPI – Network Support
  3. How do I run Open MPI over RoCE? (UCX PML)
by

Installing RoCE using Mellanox (Nvidia) OFED package

Prerequisites:

Do read Basic Understanding RoCE and Infiniband

Step 1: Install Mellanox Package

First and Foremost, you have to install Mellanox Package which you can download at https://developer.nvidia.com/networking/ethernet-software. You may want to consider installing using the traditional method or Ansible Method (Installing Mellanox OFED (mlnx_ofed) packages using Ansible)

Step 2: Load the Drivers

Activate two kernel modules that are needed for rdma and RoCE exchanges by using the command

# modprobe rdma_cm ib_umad

Step 3: Verify the drivers are loaded

# ibv_devinfo

Step 4: Set the RoCE to version 2

Set the version of the RoCE protocol to v2 by issuing the command below.

  • -d is the device, 
  • -p is the port 
  • -m the version of RoCE:
[root@node1]# cma_roce_mode -d mlx5_0 -p 1 -m 2
RoCE v2

Step 5: Check which RoCE devices are enabled on the Ethernet

[root@node-1]# ibdev2netdev
mlx5_0 port 1 ==> ens1f0 (Up)
mlx5_1 port 1 ==> ens1f1 (Down)

Refererences:

  1. Setting up a RoCE cluster
by

Basic Understanding RoCE and Infiniband

Prerequisites:

  1. RoCE required Compliant Ethernet. Currently, I am using Mellanox ConnectX-6 Cards
  2. RoCE required a Compliant Switch. I used Mellanox 100G Switch.

The Difference between Traditional Ethernet Communication and RoCE can be explained very clearly in the diagram taken by Huawei’s Basic Knowledge and Differences of RoCE, IB, and TCP Networks

Some Key Pointers on the difference between TCP/IP and RDMA

  1. The Traditional TCP/IP network communication uses the Kernel to send messages which have high data movement and data replication overhead.
  2. RDMA can bypass the kernel and access the memory directly which allows low-latency network communication.

There are 3 types of RDMA network technologies is so neatly presented in Basic Knowledge and Differences of RoCE, IB, and TCP Networks

References:

  1. Basic Knowledge and Differences of RoCE, IB, and TCP Networks