search

ina-foss / inaSpeechSegmenter

CNN-based audio segmentation toolkit. Allows to detect speech, music, noise and speaker gender. Has been designed for large scale gender equality studies based on speech time per gender.
MIT License
717 stars 127 forks source link

Cuda 11 ? #49

Closed cerw closed 3 years ago

cerw commented 3 years ago

Please fill out this template for a bug report.

Make sure you ran the unit tests before submitting an issue and tell us if and where they fail.

System information

Expected Behavior

Shoudl use GPU

Current Behavior

ina_speech_segmenter.py  -i *.mp3 -d smn -g true -o .

but it complains about missing libs, which maybe Cuda 11 does not have ? 

inaSpeechSegEnv) root@timemachine:~/radio1# ina_speech_segmenter.py  -i *.mp3 -d smn -g true -o .
2020-12-06 09:47:53.035831: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory
2020-12-06 09:47:53.035899: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
2020-12-06 09:48:00.912205: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcuda.so.1
2020-12-06 09:48:00.917185: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:982] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2020-12-06 09:48:00.918312: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1716] Found device 0 with properties:
pciBusID: 0000:00:05.0 name: GeForce GTX 1080 Ti computeCapability: 6.1
coreClock: 1.582GHz coreCount: 28 deviceMemorySize: 10.92GiB deviceMemoryBandwidth: 451.17GiB/s
2020-12-06 09:48:00.918560: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory
2020-12-06 09:48:00.919860: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcublas.so.10'; dlerror: libcublas.so.10: cannot open shared object file: No such file or directory
2020-12-06 09:48:00.949916: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcufft.so.10
2020-12-06 09:48:00.962797: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcurand.so.10
2020-12-06 09:48:00.964212: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcusolver.so.10'; dlerror: libcusolver.so.10: cannot open shared object file: No such file or directory
2020-12-06 09:48:00.964355: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcusparse.so.10'; dlerror: libcusparse.so.10: cannot open shared object file: No such file or directory
2020-12-06 09:48:00.965640: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcudnn.so.7'; dlerror: libcudnn.so.7: cannot open shared object file: No such file or directory
2020-12-06 09:48:00.965672: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1753] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...

Steps to Reproduce

  1. Install Ubuntu 18
  2. Install via PIP via virtualenv
  3. Install CUDA via Network Package
  4. Reboot
  5. Run

Additional infos

cerw commented 3 years ago

Anyone?

cerw commented 3 years ago

OK i managed to get it working on Cuda 10.1

but now i have this: 


MemoryError: Unable to allocate 280. MiB for an array with shape (179999, 1632) and data type bool
Killed```
r-uro commented 3 years ago

Hello, Sorry for not answering sooner.

For Cuda 11.1, it's probably because of a compatibility issue between Cuda and your Nvidia driver. Cuda 11.1 requires a nvidia >= 450.80.02. (https://docs.nvidia.com/deploy/cuda-compatibility/index.html)

As for the memory issue, it seems like there's not enough memory on your GPU, but 280MiB is not a lot so that's weird. Can you monitor the memory usage with nvidia-smi or nvtop when lauching inaSpeechSegmenter?

DavidDoukhan commented 3 years ago

Is this problem fixed ? Did you find any workaround ? Otherwise, we may help assuming you provid us the output of command nvidia-smi on your machine . Kind regards,

cerw commented 3 years ago

Hi Yes I have installed cuda 10.1 nad instead of running on a bunch of files I have a script that runs it on an individual file, and that seem to work.