search

pkhungurn / talking-head-anime-demo

Demo for the "Talking Head Anime from a Single Image."
MIT License
2k stars 286 forks source link

didn't work on CUDA ver. 10.2 or higher #26

Open frafrakki opened 3 years ago

frafrakki commented 3 years ago

As in title, code you've uploaded didn't work on 10.2 or higher version of CUDA/Pytorch. (This error may have occurred because of Pytorch.) BTW, below is my environment

Ryzen 7 3800X
RTX 2080 Super
RAM 24GB

Installed python

Python 3.7.9
Package           Version
----------------- ------------
backcall          0.2.0
cmake             3.18.4.post1
colorama          0.4.4
decorator         4.4.2
dlib              19.21.1
ipython           7.20.0
ipython-genutils  0.2.0
jedi              0.18.0
numpy             1.20.0
opencv-python     4.5.1.48
parso             0.8.1
pickleshare       0.7.5
Pillow            8.1.0
pip               20.1.1
prompt-toolkit    3.0.14
Pygments          2.7.4
setuptools        47.1.0
torch             1.7.1+cu101
torchaudio        0.7.2
torchvision       0.8.2+cu101
traitlets         5.0.5
typing-extensions 3.7.4.3
wcwidth           0.2.5
wheel             0.36.2

Last but not least, thanks for your great job! I'm very excited on how this project goes on.

dragonmeteor commented 3 years ago

Can you give me some error messages?

Also, have you tried creating a new environment with Anaconda?

frafrakki commented 3 years ago

Sorry, I'm not sure about the error message but I think it was something like "AssertionError: Torch not compiled with CUDA enabled". I didn't try creating environment with Anaconda. But when I've downgraded CUDA version and Pytorch, it performed absolutely perfect.

dragonmeteor commented 3 years ago

Does this help? https://github.com/pytorch/pytorch/issues/30664

frafrakki commented 3 years ago

I tried that on CUDA ver. 10.2,11.0 but nothing changed. Only 10.1 worked fine.

dragonmeteor commented 3 years ago

Here's how I created my environment:

$ conda create -n tha python=3.7
$ conda activate tha
$ conda install -c conda-forge dlib
$ pip install opencv-python
$ conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

Here's the output of conda list

# packages in environment at C:\tools\Anaconda3\envs\tha:
#
# Name                    Version                   Build  Channel
blas                      1.0                         mkl
ca-certificates           2021.1.19            haa95532_0
certifi                   2020.12.5        py37h03978a9_1    conda-forge
cudatoolkit               10.2.89              h74a9793_1
dlib                      19.21.1          py37h070da05_0    conda-forge
freetype                  2.10.4               hd328e21_0
intel-openmp              2020.3             h57928b3_311    conda-forge
jpeg                      9d                   h8ffe710_0    conda-forge
libblas                   3.9.0                     8_mkl    conda-forge
libcblas                  3.9.0                     8_mkl    conda-forge
liblapack                 3.9.0                     8_mkl    conda-forge
libpng                    1.6.37               h1d00b33_2    conda-forge
libtiff                   4.1.0                h56a325e_1
libuv                     1.40.0               he774522_0
lz4-c                     1.9.3                h2bbff1b_0
mkl                       2020.4             hb70f87d_311    conda-forge
ninja                     1.10.2           py37h6d14046_0
numpy                     1.20.1           py37hd20adf4_0    conda-forge
olefile                   0.46                     py37_0
opencv-python             4.5.1.48                 pypi_0    pypi
openssl                   1.1.1i               h8ffe710_0    conda-forge
pillow                    8.1.0            py37h4fa10fc_0
pip                       20.3.3           py37haa95532_0
python                    3.7.9                h60c2a47_0
python_abi                3.7                     1_cp37m    conda-forge
pytorch                   1.7.0           py3.7_cuda102_cudnn7_0    pytorch
setuptools                52.0.0           py37haa95532_0
sqlite                    3.33.0               h2a8f88b_0
tk                        8.6.10               he774522_0
torchaudio                0.7.0                      py37    pytorch
torchvision               0.8.1                py37_cu102    pytorch
typing_extensions         3.7.4.3            pyha847dfd_0
vc                        14.2                 h21ff451_1
vs2015_runtime            14.27.29016          h5e58377_2
wheel                     0.36.2             pyhd3eb1b0_0
wincertstore              0.2                      py37_0
xz                        5.2.5                h62dcd97_0
zlib                      1.2.11               h62dcd97_4
zstd                      1.4.5                h04227a9_0

I could run both demo programs fine. I have CUDA Toolkit v10.2.89 installed on my system.