Unable to initialize backend 'gpu'

[kuixu]

Unable to initialize backend 'gpu'

[kuixu]

Does the NVIDIA CUDA driver install successfully?

[tavolivos]

I have the same issue, even I can run it in CPU.

I installed nvidia-455 and cuda 11.1

[HGX-001]

Yes, I am sure that I have successfully installed the NVIDIA CUDA driver, because after that I installed the alphafold2 docker version given by Deep Mind, and it can run successfully, and this warning will not be reported. Before that, I also use anaconda to run other deep learning tasks and there is no problem at all. GPUs will all be used. In order to confirm the problem, I compared the same target such as T1050. The docker version of alpahafold only takes about 3 hours, and the anaconda you gave The version is well over 3 hours.

----------------

---------- Origin message ----------

From："Gustavol Enrique Olivos Ramirez" ***@***.***> To："kuixu/alphafold" ***@***.***> Subject：Re: [kuixu/alphafold] Unable to initialize backend 'gpu' (#8) Date：2021-08-26 12:34:52I have the same issue, even I can run it in CPU.

I installed nvidia-455 and cuda 11.1

…

-- You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: #8 (comment)

[JuergenUniVie]

same problem...
is it possible to set a enviroment to fix the platform "cuda" issue?

I1108 16:51:57.959832 139635007661120 xla_bridge.py:231] Unable to initialize backend 'tpu_driver': NOT_FOUND: Unable to find driver in registry given worker:
I1108 16:51:57.972085 139635007661120 xla_bridge.py:231] Unable to initialize backend 'gpu': NOT_FOUND: Could not find registered platform with name: "cuda". Available platform names are: Interpreter Host
I1108 16:51:57.972550 139635007661120 xla_bridge.py:231] Unable to initialize backend 'tpu': INVALID_ARGUMENT: TpuPlatform is not available.
W1108 16:51:57.972653 139635007661120 xla_bridge.py:236] No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 495.29.05 Driver Version: 495.29.05 CUDA Version: 11.5 |
|-------------------------------+----------------------+----------------------+
| 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 GeForce ... Off | 00000000:18:00.0 Off | N/A |
| 23% 27C P8 10W / 250W | 15MiB / 11178MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 1 NVIDIA GeForce ... Off | 00000000:3B:00.0 Off | N/A |
| 23% 28C P8 8W / 250W | 2MiB / 11178MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 2 NVIDIA GeForce ... Off | 00000000:86:00.0 Off | N/A |
| 23% 30C P8 9W / 250W | 2MiB / 11178MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
| 3 NVIDIA GeForce ... Off | 00000000:AF:00.0 Off | N/A |
| 23% 30C P8 8W / 250W | 2MiB / 11178MiB | 0% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 3626 G /usr/libexec/Xorg 9MiB |
| 0 N/A N/A 4701 G /usr/bin/gnome-shell 3MiB |
+-----------------------------------------------------------------------------+

[mattiasmar]

Hi guys,
Any insights with regards to the cuda issue?
Ping @JuergenUniVie

[deejy]

Trying to install alphafold on various platform (including AWS) since 2 months : I always fall on that bug which is frequent but not addressed.
Very sad
JPierre

[coliva92]

I'm facing the same issue. Before running the run_docker.py script I ran the docker run --rm --gpus all nvidia/cuda:11.0-base nvidia-smi command as mentioned in the README file and it correctly displayed my GPU (with no processes running). But I still got the same error messages others are reporting here.