FileNotFoundError: [Errno 2] No such file or directory: ':/usr/local/cuda-11.7/bin/nvcc'

[QiaolinLu]

Hi, I am tring to install flash-attention with pip install flash-attn==1.0.4, I meet the error, No such file or directory: ':/usr/local/cuda-11.7/bin/nvcc'

Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Collecting flash-attn==1.0.4
  Using cached https://pypi.tuna.tsinghua.edu.cn/packages/25/0f/80f16a741e328342ed4e9b3c010ec87f0ef939a71154b7b8bd22d70a0bb8/flash_attn-1.0.4.tar.gz (2.0 MB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [21 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/tmp/pip-install-u3qbej0b/flash-attn_5f7ed8b4eb05405eaeb0adf0e4265254/setup.py", line 109, in <module>
          _, bare_metal_version = get_cuda_bare_metal_version(CUDA_HOME)
        File "/tmp/pip-install-u3qbej0b/flash-attn_5f7ed8b4eb05405eaeb0adf0e4265254/setup.py", line 24, in get_cuda_bare_metal_version
          raw_output = subprocess.check_output([cuda_dir + "/bin/nvcc", "-V"], universal_newlines=True)
        File "/home/qiaolinlu/miniconda3/envs/scGPT/lib/python3.8/subprocess.py", line 415, in check_output
          return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
        File "/home/qiaolinlu/miniconda3/envs/scGPT/lib/python3.8/subprocess.py", line 493, in run
          with Popen(*popenargs, **kwargs) as process:
        File "/home/qiaolinlu/miniconda3/envs/scGPT/lib/python3.8/subprocess.py", line 858, in __init__
          self._execute_child(args, executable, preexec_fn, close_fds,
        File "/home/qiaolinlu/miniconda3/envs/scGPT/lib/python3.8/subprocess.py", line 1704, in _execute_child
          raise child_exception_type(errno_num, err_msg, err_filename)
      FileNotFoundError: [Errno 2] No such file or directory: ':/usr/local/cuda-11.7/bin/nvcc'


      torch.__version__  = 1.13.1+cu117


      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

before that, I have already installed pytorch 1.13.1+cu117, and nvcc -V is shown

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_May__3_18:49:52_PDT_2022
Cuda compilation tools, release 11.7, V11.7.64
Build cuda_11.7.r11.7/compiler.31294372_0

Could you please help me to fix it, thanks

[tridao]

Probably CUDA_HOME or PATH aren't set properly. Can you try setting CUDA_HOME to the right location such that $CUDA_HOME/bin/nvcc exists?

[QiaolinLu]

I not sure if my CUDA_HOME is right, if it was wrong, how to set it

[tridao]

You can find the full path of nvcc (e.g. with which nvcc) then set CUDA_HOME properly (e.g. export CUDA_HOME=blah).

[QiaolinLu]

I used which nvcc , result is /usr/local/cuda-11.7/bin/nvcc, in .bashrc file, it was

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.7/lib64
export PATH=$PATH:/usr/local/cuda-11.7/bin
export CUDA_HOME=$CUDA_HOME:/usr/local/cuda-11.7

[tridao]

Then I don't know what's wrong. We recommend the Pytorch container from Nvidia, which has all the required tools to install FlashAttention.

[tridao]

Looks like you set CUDA_HOME to :/usr/local/cuda-11.7 (extra : at the beginning that causes the path to be wrong).

[QiaolinLu]

Thanks a lot, I will try it.

[ll0ruc]

I have also encountered the same problem. How did you solve it @QiaolinLu

[Ash-one]

I checked my $CUDA_HOME, there was an extra : at the beginning, then I deleted it in the path, everything is fine now!

[JPZ4-5]

Looks like you set CUDA_HOME to :/usr/local/cuda-11.7 (extra : at the beginning that causes the path to be wrong).

How NICE!

That works for me since there are too many different CUDA-1X.X on my environment.

"pip install flash-attn --no-build-isolation" worked perfectly after I changed from

export CUDA_HOME=$CUDA_HOME:/path/to/CUDA-11.8

to

export CUDA_HOME=/path/to/CUDA-11.8

The possible reason for this issue could be that the program cannot correctly recognize the environment variable path containing ":".