Pip install pennylane-lightling GPU error

While installing the pennylane-lightning-gpu==0.33.0 version on the linux system, the below error occurs:


I tried to change the version of pennylane-lightning-gpu but the same error still happens.
Here is my procedure to install pennylane-gpu vesion.
step 1: pip install cuquantum-python-cu11 (installed successfully)
step 2: pip install PennyLane-Lightning-GPU==0.33.0 (failed.)

Below is the info of nvidia smi.
| NVIDIA-SMI 525.85.12 Driver Version: 525.85.12 CUDA Version: 12.0 |
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
| 0 NVIDIA A800 80G… Off | 00000000:CA:00.0 Off | Off |
| N/A 28C P0 62W / 300W | 1643MiB / 81920MiB | 0% Default |
| | | Disabled |

| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |

What should I do to solve this problem?

Thanks for your help.

Hi @Feng_Xiong, welcome to the Forum!

Our latest version available of PennyLane and pennylane-lightning-gpu is version 0.36. Are you able to update to this version? You would also need to update custatevec to use CUDA 12.

  • Firstly, you can try upgrading the lightning.gpu installed package as python -m pip install pennylane pennylane-lightning pennylane-lightning-gpu --upgrade to ensure the version is the most recent (v0.36.0 as of writing). Once this is installed and up to date, you can try pip install custatevec-cu12 to bring in the CUDA 12 variant of the NVIDIA cuQuantum custatevec library.
  • lightning.gpu requires a CUDA compute capability of SM7.0 or newer (Volta era and newer) to work. This is a hard requirement from the NVIDIA cuQuantum SDK (specifically the custatevec library), as the library expects data-center/HPC grade hardware to run. I cannot tell which GPUs you have installed, but it may be possible they are not supported by the library. You can inspect the compute capabilities for a variety of NVIDIA devices here

I hope this can help you get running in no time!

Hello Catalina,

I’ve confirmed that “custatevec-cu12” is already installed, but I’m still encountering issues when attempting to install “pennylane-lightning-gpu” using the command “python -m pip install pennylane pennylane-lightning pennylane-lightning-gpu --upgrade.” The error persists as shown in the screenshot below:

The Nvidia device I’m using is an A800. Based on my experience with successfully running VisionTransformer code using the PyTorch library, I suspect that the installation issue lies with “pennylane-lightning-gpu.”

Hi @Feng_Xiong, this looks like version conflicts somewhere.

Would you be able to create a new virtual environment following the instructions in Option 2 in the document below? Once you’ve created a new virtual environment please install PennyLane, Lightning-GPU, and custatevec as I mentioned in my previous post. Please let me know if this fixes your issue! If not you may need to use venv instead of miniconda.

PennyLane installation instructions - May 2024.pdf (82.9 KB)

Hi @Feng_Xiong

Looking at your original image, the problem appears to be the local PyPI mirror provided by your university (pip is being redirected to install from https://pypi.tuna.tsinghua.edu.cn/packages/92/f9/a5b378881af35696ad046f378ce75fd4449d01153a0b58a6d81b2c69213a/ rather than from PennyLane-Lightning-GPU · PyPI) — it is missing the precompiled LightningGPU wheels. We recommend directly using the wheels provided from PennyLane-Lightning-GPU · PyPI targeting Linux and the version of Python you need, or requesting your university to update the package to include the precompiled wheels.

Hope this helps.