Error compiling knem when installing MLNX_OFED for A100 (CUDA 11.8)


My box has a single A100 installed using a Super Micro motherboard (Supermicro MBD-X13SEI-F-O ATX Server Motherboard) with 128GB memory.
Ubuntu 22.04 is installed.

I have been struggling to install the software necessary to run cuDNN on my A100, including CUDA 11.8, as recommended in the Nvidia documentation.

I am trying to install:

./mlnxofedinstall --add-kernel-support --kernel uname -r

After it fails, the log files show a failure building knem:

configure: overriding kernel source and build dirs with /lib/modules/6.2.0-36-generic/build
configure: overriding kernel build dir with /lib/modules/6.2.0-36-generic/build
expr: syntax error: unexpected argument ‘LINUX_COMPILER’
/tmp/MLNX_OFED_LINUX-23.07- line 5688: [: -lt: unary operator expected
checking kernel.h kernel header… found in /lib/modules/6.2.0-36-generic/build
checking autoconf.h kernel header… found in /lib/modules/6.2.0-36-generic/build
checking kernel Makefile… found in /lib/modules/6.2.0-36-generic/build
configure: found Linux kernel release 6.2.16 in build tree
configure: WARNING: kernel build reports 6.2.16 instead of 6.2.0-36-generic
configure: WARNING: installing in 6.2.0-36-generic as requested
checking kernel build supports ccflags-y… no, reverting to EXTRA_CFLAGS

and later in this same log file found:

dpkg-shlibdeps: error: no dependency information found for /lib/x86_64-linux-gnu/ (used by debian/kn>
Hint: check if the library actually comes from a package.
dh_shlibdeps: error: dpkg-shlibdeps -Tdebian/knem.substvars debian/knem/opt/knem->
dh_shlibdeps: error: Aborting due to earlier error
make: *** [debian/rules:53: binary] Error 25
dpkg-buildpackage: error: debian/rules binary subprocess returned exit status 2

I really do apologize, I am fairly new to Linux (but have been a senior professional developer on Windows for many years).
Any ideas on the best way forward?

I have gone through the documentation several times, installing, uninstalling, re-installing every package along the way, and would appreciate any guidance to help me past this blockage.

Many thanks in advance,

Hi @briansboyd ,
Apologies for the delayed response.
I am checking on this with our Engineering team.