Failed to build SAM-2 and RuntimeError in infer

[flashingtt]

When I run the command

pip install -e .

There is an error:

RuntimeError:
The detected CUDA version (11.8) mismatches the version that was used to compile
PyTorch (12.1). Please make sure to use the same CUDA versions.

I try to use the command below to fix this error:

pip install --no-build-isolation -e .

Although I build SAM-2 successfully, there is another error when I run the inference image segmentation script:

RuntimeError: GET was unable to find an engine to execute this computation

Does anyone have the same problem?

[XinhanWang]

I have the same problem
RuntimeError:
The detected CUDA version (11.8) mismatches the version that was used to compile
PyTorch (12.1). Please make sure to use the same CUDA versions.

[Kiran-valetcloset]

I too have the same problem. I have installed the cuda toolkit 11.8, checked it from nvcc -V command and installed pytorch using 11.8 cuda too, but I still get the same error.

[laiwenq]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

[flashingtt]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

@laiwenq Thank you for your reply. I have tried

pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118
pip install --no-build-isolation -e .

Then I build SAM2 successfully, but failed to infer.
Is it a problem with the cuda version? My current version is 11.8, and I will upgrade to 12.1 later and try again.

[fangli333]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

@laiwenq Thank you for your reply. I have tried

pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118
pip install --no-build-isolation -e .

Then I build SAM2 successfully, but failed to infer. Is it a problem with the cuda version? My current version is 11.8, and I will upgrade to 12.1 later and try again.

did you solve it?

[laiwenq]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

@laiwenq Thank you for your reply. I have tried

pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118
pip install --no-build-isolation -e .

Then I build SAM2 successfully, but failed to infer. Is it a problem with the cuda version? My current version is 11.8, and I will upgrade to 12.1 later and try again.

Can you post what error you're seeing? I have no problem with torch==2.3.1 and cuda version 12.1, it is worth trying upgrade your cuda.

[flashingtt]

@laiwenq @fangli333 @HB16888 @Kiran-valetcloset @ChinaStark
After I upgraded my cuda to 12.1, all problems were solved. I can follow the official step-by-step installation smoothly.
Thanks to @laiwenq 's help.

[proxi666]

perhaps, we need cuda 12+ for it to run?

[aryansaurav]

You have to update gcc to version 9.3 or higher.. worked for me! worth a try.

[majnas]

            .-/+oossssoo+/-.               dev@av1 
        `:+ssssssssssssssssss+:`           ------- 
      -+ssssssssssssssssssyyssss+-         OS: Ubuntu 23.10 x86_64 
    .ossssssssssssssssssdMMMNysssso.       Host: GE76 Raider 11UH REV:1.0 
   /ssssssssssshdmmNNmmyNMMMMhssssss/      Kernel: 6.5.0-44-generic 
  +ssssssssshmydMMMMMMMNddddyssssssss+     Uptime: 15 mins 
 /sssssssshNMMMyhhyyyyhmNMMMNhssssssss/    Packages: 3012 (dpkg), 23 (snap) 
.ssssssssdMMMNhsssssssssshNMMMdssssssss.   Shell: zsh 5.9 
+sssshhhyNMMNyssssssssssssyNMMMysssssss+   Resolution: 2048x1152, 1920x1080, 2560x1440 
ossyNMMMNyMMhsssssssssssssshmmmhssssssso   DE: GNOME 45.2 
ossyNMMMNyMMhsssssssssssssshmmmhssssssso   WM: Mutter 
+sssshhhyNMMNyssssssssssssyNMMMysssssss+   WM Theme: Adwaita 
.ssssssssdMMMNhsssssssssshNMMMdssssssss.   Theme: Yaru-viridian [GTK2/3] 
 /sssssssshNMMMyhhyyyyhdNMMMNhssssssss/    Icons: Yaru-viridian [GTK2/3] 
  +sssssssssdmydMMMMMMMMddddyssssssss+     Terminal: gnome-terminal 
   /ssssssssssshdmNNNNmyNMMMMhssssss/      CPU: 11th Gen Intel i9-11980HK (16) @ 4.900GHz 
    .ossssssssssssssssssdMMMNysssso.       GPU: NVIDIA GeForce RTX 3080 Mobile / Max-Q 8GB/16GB 
      -+sssssssssssssssssyyyssss+-         GPU: Intel TigerLake-H GT1 [UHD Graphics] 
        `:+ssssssssssssssssss+:`           Memory: 3764MiB / 31787MiB 
            .-/+oossssoo+/-.
                                                                   
                                                                   

I have tested gcc-9/10/11 none of them worked. finally upgrade cuda toolkit to 12.1 solved the problem

wget https://developer.download.nvidia.com/compute/cuda/12.1.0/local_installers/cuda_12.1.0_530.30.02_linux.run
sudo sh cuda_12.1.0_530.30.02_linux.run --toolkit --toolkitpath=/usr/local/cuda-12.1 --override

pip list
Package                  Version      Editable project location
------------------------ ------------ ---------------------------------------------------------
antlr4-python3-runtime   4.9.3
asttokens                2.4.1
comm                     0.2.2
contourpy                1.2.1
cycler                   0.12.1
debugpy                  1.8.2
decorator                5.1.1
executing                2.0.1
filelock                 3.13.1
fonttools                4.53.1
fsspec                   2024.2.0
hydra-core               1.3.2
iopath                   0.1.10
ipykernel                6.29.5
ipython                  8.26.0
jedi                     0.19.1
Jinja2                   3.1.3
jupyter_client           8.6.2
jupyter_core             5.7.2
kiwisolver               1.4.5
MarkupSafe               2.1.5
matplotlib               3.9.1
matplotlib-inline        0.1.7
mpmath                   1.3.0
nest-asyncio             1.6.0
networkx                 3.2.1
numpy                    1.26.3
nvidia-cublas-cu11       11.11.3.6
nvidia-cuda-cupti-cu11   11.8.87
nvidia-cuda-nvrtc-cu11   11.8.89
nvidia-cuda-runtime-cu11 11.8.89
nvidia-cudnn-cu11        8.7.0.84
nvidia-cufft-cu11        10.9.0.58
nvidia-curand-cu11       10.3.0.86
nvidia-cusolver-cu11     11.4.1.48
nvidia-cusparse-cu11     11.7.5.86
nvidia-nccl-cu11         2.20.5
nvidia-nvtx-cu11         11.8.86
omegaconf                2.3.0
opencv-python            4.10.0.84
packaging                24.1
parso                    0.8.4
pexpect                  4.9.0
pillow                   10.2.0
pip                      24.0
platformdirs             4.2.2
portalocker              2.10.1
prompt_toolkit           3.0.47
psutil                   6.0.0
ptyprocess               0.7.0
pure_eval                0.2.3
Pygments                 2.18.0
pyparsing                3.1.2
python-dateutil          2.9.0.post0
PyYAML                   6.0.1
pyzmq                    26.0.3
SAM-2                    1.0          /home/dev/Documents/other_repositories/segment-anything-2
setuptools               69.1.0
six                      1.16.0
stack-data               0.6.3
sympy                    1.12
torch                    2.3.1+cu118
torchaudio               2.3.1+cu118
torchvision              0.18.1+cu118
tornado                  6.4.1
tqdm                     4.66.4
traitlets                5.14.3
triton                   2.3.1
typing_extensions        4.9.0
wcwidth                  0.2.13
wheel                    0.42.0

[aryansaurav]

Indeed, so it seems the basic required config are something like:
CUDA >12.1
PyTorch 2.4 (latest would be better, but earlier versions might work too)
gcc (9.3 or higher)

of course, this is assuming a GPU machine.

if you have one or more missing in the above list, worth trying to fix that first

[YichiZhang98]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

Thank you for your reply. I have tried but failed to build.

The error shows as follows.

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building editable for SAM-2
Failed to build SAM-2
ERROR: ERROR: Failed to build installable wheels for some pyproject.toml based projects (SAM-2)

My cuda version 12.1 bu it still not work.

[Yan-0-0-Ye]

您可以将 cuda 升级到 12.1。否则手动安装依赖项，例如，pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

感谢您的回复。我试过但未能构建。

错误如下所示。

注意：此错误源自子进程，可能不是 pip 的问题。错误：无法为 SAM-2 构建可编辑内容 无法生成 SAM-2 错误：错误：无法为某些基于 pyproject.toml 的项目 （SAM-2） 构建可安装轮子

我的 cuda 版本 12.1 bu 它仍然不起作用。

hello, do you solve it?

[laiwenq]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

Thank you for your reply. I have tried but failed to build.

The error shows as follows.

note: This error originates from a subprocess, and is likely not a problem with pip. ERROR: Failed building editable for SAM-2 Failed to build SAM-2 ERROR: ERROR: Failed to build installable wheels for some pyproject.toml based projects (SAM-2)

My cuda version 12.1 bu it still not work.

Sorry just see this now. Check your cuda version by running nvcc -V in the command line and see if you are actually using 12.1.

[Yan-0-0-Ye]

You can upgrade your cuda to 12.1. Otherwise manually install the dependencies, e.g., pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu118

Thank you for your reply. I have tried but failed to build.
The error shows as follows.
note: This error originates from a subprocess, and is likely not a problem with pip. ERROR: Failed building editable for SAM-2 Failed to build SAM-2 ERROR: ERROR: Failed to build installable wheels for some pyproject.toml based projects (SAM-2)
My cuda version 12.1 bu it still not work.

Sorry just see this now. Check your cuda version by running nvcc -V in the command line and see if you are actually using 12.1.

thanks!

[ronghanghu]

Hi, we have recently made the CUDA extension step optional (in https://github.com/facebookresearch/segment-anything-2/pull/155) as a workaround to this problem.

You can pull the latest code and reinstall via

# run the line below inside the SAM 2 repo
git pull;
pip uninstall -y SAM-2;
rm -f sam2/*.so;
pip install -e ".[demo]"

which allows using SAM 2 without CUDA extension (the results should stay the same in most cases, see INSTALL.md for details).