Daily Dev Tips

When bitsandbytes Fails to Install on WSL Ubuntu

Qualiteg Product Development Team

Qualiteg Product Development Team

— 5 min read
When bitsandbytes Fails to Install on WSL Ubuntu
Photo by Gabriel Heinzer / Unsplash

This post explains how to deal with the following error when you try to pip install bitsandbytes.

Could not load bitsandbytes native library: libcusparse.so.11: cannot open shared object file: No such file or directory
Traceback (most recent call last):
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/cextension.py", line 109, in <module>
    lib = get_native_library()
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/cextension.py", line 96, in get_native_library
    dll = ct.cdll.LoadLibrary(str(binary_path))
  File "/usr/lib/python3.10/ctypes/__init__.py", line 452, in LoadLibrary
    return self._dlltype(name)
  File "/usr/lib/python3.10/ctypes/__init__.py", line 374, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: libcusparse.so.11: cannot open shared object file: No such file or directory

CUDA Setup failed despite CUDA being available. Please run the following command to get more information:

python -m bitsandbytes

Inspect the output of the command and see if you can locate CUDA libraries. You might need to add them
to your LD_LIBRARY_PATH. If you suspect a bug, please take the information from python -m bitsandbytes
and open an issue at: https://github.com/TimDettmers/bitsandbytes/issues

Displaying and Reading the Debug Information

Using Python's module execution feature, let's run bitsandbytes as a module to display its debug information.

python -m bitsandbytes

Could not load bitsandbytes native library: libcusparse.so.11: cannot open shared object file: No such file or directory
Traceback (most recent call last):
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/cextension.py", line 109, in <module>
    lib = get_native_library()
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/cextension.py", line 96, in get_native_library
    dll = ct.cdll.LoadLibrary(str(binary_path))
  File "/usr/lib/python3.10/ctypes/__init__.py", line 452, in LoadLibrary
    return self._dlltype(name)
  File "/usr/lib/python3.10/ctypes/__init__.py", line 374, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: libcusparse.so.11: cannot open shared object file: No such file or directory

CUDA Setup failed despite CUDA being available. Please run the following command to get more information:

python -m bitsandbytes

Inspect the output of the command and see if you can locate CUDA libraries. You might need to add them
to your LD_LIBRARY_PATH. If you suspect a bug, please take the information from python -m bitsandbytes
and open an issue at: https://github.com/TimDettmers/bitsandbytes/issues
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++ BUG REPORT INFORMATION ++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++ OTHER +++++++++++++++++++++++++++
CUDA specs: CUDASpecs(highest_compute_capability=(8, 6), cuda_version_string='117', cuda_version_tuple=(11, 7))
PyTorch settings found: CUDA_VERSION=117, Highest Compute Capability: (8, 6).
To manually override the PyTorch CUDA version please see: https://github.com/TimDettmers/bitsandbytes/blob/main/docs/source/nonpytorchcuda.mdx
CUDA SETUP: WARNING! CUDA runtime files not found in any environmental path.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++ DEBUG INFO END ++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Checking that the library is importable and CUDA is callable...
Couldn't load the bitsandbytes library, likely due to missing binaries.
Please ensure bitsandbytes is properly installed.

For source installations, compile the binaries with `cmake -DCOMPUTE_BACKEND=cuda -S .`.
See the documentation for more details if needed.

Trying a simple check anyway, but this will likely fail...
Traceback (most recent call last):
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/diagnostics/main.py", line 66, in main
    sanity_check()
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/diagnostics/main.py", line 40, in sanity_check
    adam.step()
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/torch/optim/optimizer.py", line 280, in wrapper
    out = func(*args, **kwargs)
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/optim/optimizer.py", line 287, in step
    self.update_step(group, p, gindex, pindex)
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/optim/optimizer.py", line 496, in update_step
    F.optimizer_update_32bit(
  File "/home/mlu/.virtualenvs/ChatStream/lib/python3.10/site-packages/bitsandbytes/functional.py", line 1584, in optimizer_update_32bit
    optim_func = str2optimizer32bit[optimizer_name][0]
NameError: name 'str2optimizer32bit' is not defined
Above we output some debug information.
Please provide this info when creating an issue via https://github.com/TimDettmers/bitsandbytes/issues/new/choose
WARNING: Please be sure to sanitize sensitive info from the output before posting it.

Along with the runtime error, a BUG REPORT INFORMATION section is displayed.

It contains the following two important pieces of information:

  • It may be a good idea to add the CUDA path to LD_LIBRARY_PATH.
  • The CUDA version used by the PyTorch installed in the current venv is 11.7, since PyTorch settings found: CUDA_VERSION=117 is displayed.

Checking the CUDA Version Installed in the Current WSL Ubuntu

Let's check which CUDA version is installed in the current WSL2 Ubuntu.

ls /usr/local/cuda* -d

Listing the CUDA installation directories with the command above shows

/usr/local/cuda /usr/local/cuda-12 /usr/local/cuda-12.4

, which tells us that CUDA version 12.4 is installed on this Ubuntu.

If you want to see more detailed version numbers on a per-module basis, you can display them with

cat /usr/local/cuda/version.json

.

Cause of the Error Message

Now the cause of the error message becomes clear.

The PyTorch installed in the venv was built for CUDA_VERSION=117, whereas the CUDA version actually installed on Ubuntu was 12.4. As a result, bitsandbytes went looking for 11.7 and could not find the CUDA libraries.

Solutions

There are three possible solutions:

  1. Change the CUDA libraries on Ubuntu to CUDA 11.7, matching the venv
  2. Change the libraries in the venv to CUDA 12.4, matching Ubuntu
  3. Tell bitsandbytes to use CUDA 12.4

Here we will go with option 3, which is the easiest.

(If the application's tests pass with this, there is no problem. If they do not, you will need to fall back on option 1 or 2.)

Telling bitsandbytes the CUDA Version

To point bitsandbytes at the CUDA 12.4 libraries, do the following:

export BNB_CUDA_VERSION=124

Now let's run bitsandbytes as a module once again.

python -m bitsandbytes

WARNING: BNB_CUDA_VERSION=124 environment variable detected; loading libbitsandbytes_cuda124.so.
This can be used to load a bitsandbytes version that is different from the PyTorch CUDA version.
If this was unintended set the BNB_CUDA_VERSION variable to an empty string: export BNB_CUDA_VERSION=
If you use the manual override make sure the right libcudart.so is in your LD_LIBRARY_PATH
For example by adding the following to your .bashrc: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:<path_to_cuda_dir/lib64

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++ BUG REPORT INFORMATION ++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++ OTHER +++++++++++++++++++++++++++
CUDA specs: CUDASpecs(highest_compute_capability=(8, 6), cuda_version_string='117', cuda_version_tuple=(11, 7))
PyTorch settings found: CUDA_VERSION=117, Highest Compute Capability: (8, 6).
WARNING: BNB_CUDA_VERSION=124 environment variable detected; loading libbitsandbytes_cuda124.so.
This can be used to load a bitsandbytes version that is different from the PyTorch CUDA version.
If this was unintended set the BNB_CUDA_VERSION variable to an empty string: export BNB_CUDA_VERSION=
If you use the manual override make sure the right libcudart.so is in your LD_LIBRARY_PATH
For example by adding the following to your .bashrc: export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:<path_to_cuda_dir/lib64

To manually override the PyTorch CUDA version please see: https://github.com/TimDettmers/bitsandbytes/blob/main/docs/source/nonpytorchcuda.mdx
CUDA SETUP: WARNING! CUDA runtime files not found in any environmental path.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++ DEBUG INFO END ++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Checking that the library is importable and CUDA is callable...
SUCCESS!
Installation was successful!

Installation was successful!

This time, bitsandbytes was able to find the CUDA libraries successfully, and it looks like it will work fine.

Making the Environment Variable Take Effect

Let's make sure BNB_CUDA_VERSION takes effect.

Setting it in .bashrc

For running Python applications from the terminal, add the BNB_CUDA_VERSION environment variable to .bashrc.

Open .bashrc and, as the last line, add

export BNB_CUDA_VERSION=124

, then save the file.

Setting it directly in code

You can also set it directly from Python code as follows:

import os

# Set the BNB_CUDA_VERSION environment variable
os.environ['BNB_CUDA_VERSION'] = '124'

Read more