Skip to content

RHEL/CentOS version can't find libnvidia-ml #16

@kcgthb

Description

@kcgthb

Hi!

It looks like the RHEL/CentOS binary version looks for libnvidia-ml.so whereas the NVML library that is provided by the nvidia-driver-NVML RPM is libnvidia-ml.so.1:

# rpm -ql nvidia-driver-NVML-410.79-1.el7.x86_64
/usr/lib64/libnvidia-ml.so.1
/usr/lib64/libnvidia-ml.so.410.79

# ./ngputop_v0.1.4_Ubuntu_x64
NVML library is not installed or not found at the default path.
NVIDIA driver not found.

# strace -e open ./ngputop_v0.1.4_Ubuntu_x64 |& grep libnvidia-ml
open("/lib64/tls/x86_64/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/lib64/tls/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/lib64/x86_64/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/lib64/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/lib64/tls/x86_64/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/lib64/tls/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/lib64/x86_64/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
open("/usr/lib64/libnvidia-ml.so", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)

Linking libnvidia-ml.so.1 to libnvidia-ml.so does resolve the problem, but it would probably be better if ngputop could make use of the libraries provided by the NVML RPMs directly.

Thanks!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions