HPC-X General Support

NVIDIA HPC-X Software Toolkit Rev 2.17.1

The platform and requirements for HPC-X are detailed in the following table:

Platform

Versions

CUDA

12.x

GDRCopy

2.3

MLNX_OFED

23.10

NCCL

2.x

NVIDIA BlueField-2

24.38.1002

NVIDIA BlueField-3

32.38.1002

NVIDIA ConnectX-5/ConnectX-5 Ex

16.35.2000

NVIDIA ConnectX-6

20.38.1900

NVIDIA ConnectX-6 Dx

22.38.1900

NVIDIA ConnectX-6 Lx

26.38.1900

NVIDIA ConnectX-7

28.38.1900

XPMEM

2.7

The following communications libraries and acceleration packages are part of this NVIDIA HPC-X® package:

Library/Acceleration Package

Version Number

Open MPI

4.1

NVIDIA Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)

3.5.0

HCOLL

4.8

UCX

1.16.0

UCC

1.3

Open SHMEM specification compliance

1.41

ClusterKit2

1.11

nccl-rdma-sharp-plugin3

2.5

  1. Full Open SHMEM v1.4 support is available only if compiled with C11 Standard (see Rebuilding Open MPI from HPC-X™ Sources).

  2. ClusterKit is a multifaceted node assessment tool for high performance clusters.

  3. nccl-rdma-sharp plugin enables RDMA and Switch-based collectives (SHARP) with NVIDIA's NCCL library.

Warning

When HPC-X is launched with Open MPI without a resource manager job environment (slurm,pbs, etc.), or when it is launched from a compute node, the default rsh/ssh-based launcher will be used. This launcher does not propagate environment variables to the compute nodes. Thus, it is important to ensure the propagation of LD_LIBRARY_PATH variable from HPC-x is done as follows.

Copy
Copied!
            

%mpirun -x LD_LIBRARY_PATH -np 2 -H host1,host2 $HPCX_MPI_TESTS_DIR/examples/hello_c

The following table lists the supported operating systems and CPUs for the latest HPC-X.

Warning

Starting from HPC-X v2.9, HPC-X will no longer support PPC architecture.

Operating System

Platforms

RHEL/CentOS/Rocky 7.x

x86_64, aarch64

RHEL/CentOS/Rocky 8.x

x86_64, aarch64

RHEL/CentOS/Rocky 9.x

x86_64, aarch64

CentOS 8.x Stream

x86_64

CentOS 9.x Stream

x86_64

SLES 12 SP4

x86_64, aarch64

SLES 12 SP5

x86_64

SLES 15 SP2

x86_64

SLES 15 SP3

x86_64

SLES 15 SP4

x86_64

Ubuntu 18.04

x86_64

Ubuntu 20.04

x86_64, aarch64

Ubuntu 22.04

x86_64, aarch64

OpenEuler 20.03

x86_64, aarch64

Kylin 10 SP1

x86_64, aarch64

Kylin 10 SP2

x86_64, aarch64

Debian 10.x

x86_64

Debian 11.x

x86_64

© Copyright 2023, NVIDIA. Last updated on Dec 12, 2023.