Unable to install flash_attn on H100 with CUDA 12.5

[ghadiaravi13]

Hi,

I am trying to install flash_attn on H100 with CUDA driver 12.6, and I already installed torch with cu12.6 nightly

torch.cuda.is_available() = True on my system

However, I get the below error when trying to install flash_attn:

nvc++-Error-Unknown switch: -Wno-unused-result
nvc++-Error-Unknown switch: -grecord-gcc-switches
nvc++-Error-Unknown switch: -pipe
nvc++-Error-Unknown switch: -Wp,-D_FORTIFY_SOURCE=2
nvc++-Error-Unknown switch: -Wp,-D_GLIBCXX_ASSERTIONS
nvc++-Error-Unknown switch: -fstack-protector-strong
nvc++-Error-Unknown switch: -fasynchronous-unwind-tables
nvc++-Error-Unknown switch: -fstack-clash-protection
nvc++-Error-Unknown switch: -fwrapv
nvc++-Error-Unknown switch: -grecord-gcc-switches
nvc++-Error-Unknown switch: -pipe
nvc++-Error-Unknown switch: -Wp,-D_FORTIFY_SOURCE=2
nvc++-Error-Unknown switch: -Wp,-D_GLIBCXX_ASSERTIONS
nvc++-Error-Unknown switch: -fstack-protector-strong
nvc++-Error-Unknown switch: -fasynchronous-unwind-tables
nvc++-Error-Unknown switch: -fstack-clash-protection
nvc++-Error-Unknown switch: -fwrapv
nvc++-Error-Unknown switch: -grecord-gcc-switches
nvc++-Error-Unknown switch: -pipe
nvc++-Error-Unknown switch: -Wp,-D_FORTIFY_SOURCE=2
nvc++-Error-Unknown switch: -Wp,-D_GLIBCXX_ASSERTIONS
nvc++-Error-Unknown switch: -fstack-protector-strong
nvc++-Error-Unknown switch: -fasynchronous-unwind-tables
nvc++-Error-Unknown switch: -fstack-clash-protection
nvc++-Error-Unknown switch: -fwrapv
error: command '/home1/apps/nvidia/Linux_aarch64/24.7/compilers/bin/nvc++' failed with exit code 1
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for flash-attn
Running setup.py clean for flash-attn
Failed to build flash-attn

Can someone help suggest if they have faced similar error earlier?