Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

MKL Version instability #42

Open
DDXDB opened this issue Jan 1, 2024 · 19 comments
Open

MKL Version instability #42

DDXDB opened this issue Jan 1, 2024 · 19 comments
Labels
bug Something isn't working

Comments

@DDXDB
Copy link

DDXDB commented Jan 1, 2024 •
edited by CarlGao4
Loading

happen

Sometimes the GPU is not recognized at startup, and only the CPU is optional in the device bar
The startup task crashes after selecting the GPU

GPU：A770 16G（[Acer]）
Driver：31.0.101.5084

What is the platform you are using?

Windows

log

2024-01-01 23:50:46,836 (GuiMain.py) (Line 1412) [INFO] : Python version: 3.11.7 | packaged by conda-forge | (main, Dec 23 2023, 14:27:59) [MSC v.1937 64 bit (AMD64)]
2024-01-01 23:50:46,836 (GuiMain.py) (Line 1413) [INFO] : Demucs GUI version: 1.1a2
2024-01-01 23:50:46,882 (GuiMain.py) (Line 1414) [INFO] : System: Windows-10-10.0.22621-SP0
2024-01-01 23:50:46,882 (GuiMain.py) (Line 1415) [INFO] : Architecture: 64bit
2024-01-01 23:50:46,882 (GuiMain.py) (Line 1416) [INFO] : CPU: AMD64 Family 25 Model 33 Stepping 0, AuthenticAMD
2024-01-01 23:50:46,882 (GuiMain.py) (Line 1417) [INFO] : CPU count: 12
2024-01-01 23:50:46,889 (GuiMain.py) (Line 1418) [INFO] : System memory: 34266697728 (31.913GB)
2024-01-01 23:50:46,894 (GuiMain.py) (Line 1421) [INFO] : System free memory: 23235440640 (21.64GB)
2024-01-01 23:50:47,303 (GuiMain.py) (Line 1424) [INFO] : System swap memory: 38654705664 (36GB)
2024-01-01 23:50:47,303 (GuiMain.py) (Line 1435) [INFO] : Using PySide6
2024-01-01 23:50:47,303 (GuiMain.py) (Line 1436) [INFO] : Qt version: 6.6.1
2024-01-01 23:50:47,303 (GuiMain.py) (Line 1437) [INFO] : PySide6 version: 6.6.1
2024-01-01 23:50:47,652 (shared.py) (Line 243) [INFO] : [1] Thread starter (separator.py) starts
2024-01-01 23:50:47,716 (GuiMain.py) (Line 1442) [DEBUG] : Supported styles: windowsvista, Windows, Fusion
2024-01-01 23:50:47,718 (GuiMain.py) (Line 1446) [DEBUG] : Current style: windowsvista
2024-01-01 23:50:52,464 (audio.py) (Line 34) [INFO] : Soundfile version: 0.12.1
2024-01-01 23:50:52,464 (audio.py) (Line 35) [INFO] : libsndfile version: 1.2.0
2024-01-01 23:50:52,464 (audio.py) (Line 36) [INFO] : SoXR version: 0.3.7
2024-01-01 23:50:52,464 (audio.py) (Line 37) [INFO] : libsoxr version: 0.1.3-7-gacc0dac
2024-01-01 23:50:52,465 (separator.py) (Line 62) [INFO] : Demucs version: 4.1.0a2
2024-01-01 23:50:52,465 (separator.py) (Line 63) [INFO] : PyTorch version: 2.1.0a0+cxx11.abi
2024-01-01 23:50:53,507 (audio.py) (Line 49) [INFO] : ffmpeg -version output:
ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil      58.  2.100 / 58.  2.100
libavcodec     60.  3.100 / 60.  3.100
libavformat    60.  3.100 / 60.  3.100
libavdevice    60.  1.100 / 60.  1.100
libavfilter     9.  3.100 /  9.  3.100
libswscale      7.  1.100 /  7.  1.100
libswresample   4. 10.100 /  4. 10.100
libpostproc    57.  1.100 / 57.  1.100

2024-01-01 23:50:53,550 (audio.py) (Line 54) [INFO] : Using ffmpeg from E:\Downloads\Demucs-GUI_1.1a2_mkl\ffmpeg\ffmpeg.EXE
2024-01-01 23:50:53,551 (audio.py) (Line 55) [INFO] : ffprobe -version output:
ffprobe version 6.0-full_build-www.gyan.dev Copyright (c) 2007-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil      58.  2.100 / 58.  2.100
libavcodec     60.  3.100 / 60.  3.100
libavformat    60.  3.100 / 60.  3.100
libavdevice    60.  1.100 / 60.  1.100
libavfilter     9.  3.100 /  9.  3.100
libswscale      7.  1.100 /  7.  1.100
libswresample   4. 10.100 /  4. 10.100
libpostproc    57.  1.100 / 57.  1.100

2024-01-01 23:50:54,551 (shared.py) (Line 254) [INFO] : [1] Thread starter (separator.py) ends
2024-01-01 23:50:54,554 (shared.py) (Line 243) [INFO] : [2] Thread checkUpdate (shared.py) starts
2024-01-01 23:50:54,564 (shared.py) (Line 270) [INFO] : Checking for updates...
2024-01-01 23:50:56,460 (shared.py) (Line 273) [INFO] : Latest version: 1.1a2
2024-01-01 23:50:56,461 (shared.py) (Line 254) [INFO] : [2] Thread checkUpdate (shared.py) ends
2024-01-01 23:51:32,988 (shared.py) (Line 243) [INFO] : [3] Thread loadModel (GuiMain.py) starts
2024-01-01 23:51:32,991 (GuiMain.py) (Line 505) [INFO] : Loading model htdemucs from repo E:\Downloads\Demucs-GUI_1.1a2_mkl\pretrained
2024-01-01 23:51:33,363 (GuiMain.py) (Line 527) [INFO] : Model loaded within 0.3696s
2024-01-01 23:51:33,371 (GuiMain.py) (Line 528) [INFO] : Model: htdemucs
Repo: E:\Downloads\Demucs-GUI_1.1a2_mkl\pretrained
Type: Bag of models
Audio channels: 2
Sample rate: 44100
Sources: drums, bass, other, vocals

Model 0:
　Type: HTDemucs
　Default segment: 7.8
　Weight: [1.0, 1.0, 1.0, 1.0]
2024-01-01 23:51:33,373 (shared.py) (Line 254) [INFO] : [3] Thread loadModel (GuiMain.py) ends
2024-01-01 23:51:45,674 (shared.py) (Line 243) [INFO] : [4] Thread separate (separator.py) starts
2024-01-01 23:51:45,674 (separator.py) (Line 410) [INFO] : Start separating audio: 1.mp3
2024-01-01 23:51:45,674 (separator.py) (Line 411) [INFO] : Parameters: segment=7.80 overlap=0.25 shifts=0
2024-01-01 23:51:45,674 (separator.py) (Line 412) [INFO] : Device: xpu:0
2024-01-01 23:51:45,675 (audio.py) (Line 66) [DEBUG] : Reading audio with soundfile: E:\Downloads\Demucs-GUI_1.1a2_mkl\1.mp3
2024-01-01 23:51:46,073 (audio.py) (Line 82) [INFO] : Read audio E:\Downloads\Demucs-GUI_1.1a2_mkl\1.mp3: samplerate=48000 shape=(24686592, 2)
2024-01-01 23:51:46,073 (audio.py) (Line 85) [INFO] : Samplerate 48000 doesn't match target 44100, resampling with SoXR
2024-01-01 23:51:47,387 (separator.py) (Line 442) [INFO] : Running separation...
@DDXDB DDXDB added the bug Something isn't working label Jan 1, 2024
@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 1, 2024

First of all, according to statistics, the top-3 download countries are Japan, Italy and America, so I'd like to require that all issues and pull requests are to be opened in English.

Besides, I forgot to log Intel MKL info in the logs in the previous build. I've updated the released file again and can you upload your logs again?

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 1, 2024

Related to intel/intel-extension-for-pytorch#479

@DDXDB
Copy link
Author

DDXDB commented Jan 1, 2024 •
edited
Loading

new Log

2024-01-02 00:40:33,227 (GuiMain.py) (Line 1412) [INFO] : Python version: 3.11.7 | packaged by conda-forge | (main, Dec 23 2023, 14:27:59) [MSC v.1937 64 bit (AMD64)]
2024-01-02 00:40:33,228 (GuiMain.py) (Line 1413) [INFO] : Demucs GUI version: 1.1a2
2024-01-02 00:40:33,274 (GuiMain.py) (Line 1414) [INFO] : System: Windows-10-10.0.22621-SP0
2024-01-02 00:40:33,274 (GuiMain.py) (Line 1415) [INFO] : Architecture: 64bit
2024-01-02 00:40:33,274 (GuiMain.py) (Line 1416) [INFO] : CPU: AMD64 Family 25 Model 33 Stepping 0, AuthenticAMD
2024-01-02 00:40:33,274 (GuiMain.py) (Line 1417) [INFO] : CPU count: 12
2024-01-02 00:40:33,283 (GuiMain.py) (Line 1418) [INFO] : System memory: 34266697728 (31.913GB)
2024-01-02 00:40:33,289 (GuiMain.py) (Line 1421) [INFO] : System free memory: 22504267776 (20.959GB)
2024-01-02 00:40:33,672 (GuiMain.py) (Line 1424) [INFO] : System swap memory: 38654705664 (36GB)
2024-01-02 00:40:33,672 (GuiMain.py) (Line 1435) [INFO] : Using PySide6
2024-01-02 00:40:33,672 (GuiMain.py) (Line 1436) [INFO] : Qt version: 6.6.1
2024-01-02 00:40:33,672 (GuiMain.py) (Line 1437) [INFO] : PySide6 version: 6.6.1
2024-01-02 00:40:33,953 (shared.py) (Line 243) [INFO] : [1] Thread starter (separator.py) starts
2024-01-02 00:40:34,022 (GuiMain.py) (Line 1442) [DEBUG] : Supported styles: windowsvista, Windows, Fusion
2024-01-02 00:40:34,025 (GuiMain.py) (Line 1446) [DEBUG] : Current style: windowsvista
2024-01-02 00:40:41,674 (separator.py) (Line 53) [INFO] : Intel Extension for PyTorch version: 2.1.10+xpu
2024-01-02 00:40:41,757 (audio.py) (Line 34) [INFO] : Soundfile version: 0.12.1
2024-01-02 00:40:41,757 (audio.py) (Line 35) [INFO] : libsndfile version: 1.2.0
2024-01-02 00:40:41,757 (audio.py) (Line 36) [INFO] : SoXR version: 0.3.7
2024-01-02 00:40:41,757 (audio.py) (Line 37) [INFO] : libsoxr version: 0.1.3-7-gacc0dac
2024-01-02 00:40:41,757 (separator.py) (Line 63) [INFO] : Demucs version: 4.1.0a2
2024-01-02 00:40:41,757 (separator.py) (Line 64) [INFO] : PyTorch version: 2.1.0a0+cxx11.abi
2024-01-02 00:40:41,758 (separator.py) (Line 79) [INFO] : Intel MKL Info: _DeviceProperties(name='Intel(R) Arc(TM) A770 Graphics', platform_name='Intel(R) Level-Zero', dev_type='gpu, support_fp64=0, total_memory=15930MB, max_compute_units=512, gpu_eu_count=512)
2024-01-02 00:40:42,807 (audio.py) (Line 49) [INFO] : ffmpeg -version output:
ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil 58. 2.100 / 58. 2.100
libavcodec 60. 3.100 / 60. 3.100
libavformat 60. 3.100 / 60. 3.100
libavdevice 60. 1.100 / 60. 1.100
libavfilter 9. 3.100 / 9. 3.100
libswscale 7. 1.100 / 7. 1.100
libswresample 4. 10.100 / 4. 10.100
libpostproc 57. 1.100 / 57. 1.100

2024-01-02 00:40:42,852 (audio.py) (Line 54) [INFO] : Using ffmpeg from D:\Program Files\Demucs-GUI_1.1a2_mkl\ffmpeg\ffmpeg.EXE
2024-01-02 00:40:42,852 (audio.py) (Line 55) [INFO] : ffprobe -version output:
ffprobe version 6.0-full_build-www.gyan.dev Copyright (c) 2007-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil 58. 2.100 / 58. 2.100
libavcodec 60. 3.100 / 60. 3.100
libavformat 60. 3.100 / 60. 3.100
libavdevice 60. 1.100 / 60. 1.100
libavfilter 9. 3.100 / 9. 3.100
libswscale 7. 1.100 / 7. 1.100
libswresample 4. 10.100 / 4. 10.100
libpostproc 57. 1.100 / 57. 1.100

2024-01-02 00:40:43,853 (shared.py) (Line 254) [INFO] : [1] Thread starter (separator.py) ends
2024-01-02 00:40:43,857 (shared.py) (Line 243) [INFO] : [2] Thread checkUpdate (shared.py) starts
2024-01-02 00:40:43,870 (shared.py) (Line 270) [INFO] : Checking for updates...
2024-01-02 00:40:44,783 (shared.py) (Line 273) [INFO] : Latest version: 1.1a2
2024-01-02 00:40:44,783 (shared.py) (Line 254) [INFO] : [2] Thread checkUpdate (shared.py) ends
2024-01-02 00:40:45,475 (shared.py) (Line 243) [INFO] : [3] Thread loadModel (GuiMain.py) starts
2024-01-02 00:40:45,477 (GuiMain.py) (Line 505) [INFO] : Loading model htdemucs from repo D:\Program Files\Demucs-GUI_1.1a2_mkl\pretrained
2024-01-02 00:40:45,819 (GuiMain.py) (Line 527) [INFO] : Model loaded within 0.3384s
2024-01-02 00:40:45,819 (GuiMain.py) (Line 528) [INFO] : Model: htdemucs
Repo: D:\Program Files\Demucs-GUI_1.1a2_mkl\pretrained
Type: Bag of models
Audio channels: 2
Sample rate: 44100
Sources: drums, bass, other, vocals

Model 0:
　Type: HTDemucs
　Default segment: 7.8
　Weight: [1.0, 1.0, 1.0, 1.0]
2024-01-02 00:40:45,819 (shared.py) (Line 254) [INFO] : [3] Thread loadModel (GuiMain.py) ends
2024-01-02 00:41:03,371 (shared.py) (Line 243) [INFO] : [4] Thread separate (separator.py) starts
2024-01-02 00:41:03,371 (separator.py) (Line 415) [INFO] : Start separating audio: 1.mp3
2024-01-02 00:41:03,371 (separator.py) (Line 416) [INFO] : Parameters: segment=7.80 overlap=0.25 shifts=0
2024-01-02 00:41:03,371 (separator.py) (Line 417) [INFO] : Device: xpu:0
2024-01-02 00:41:03,371 (audio.py) (Line 66) [DEBUG] : Reading audio with soundfile: D:\Program Files\Demucs-GUI_1.1a2_mkl\1.mp3
2024-01-02 00:41:03,796 (audio.py) (Line 82) [INFO] : Read audio D:\Program Files\Demucs-GUI_1.1a2_mkl\1.mp3: samplerate=48000 shape=(24686592, 2)
2024-01-02 00:41:03,796 (audio.py) (Line 85) [INFO] : Samplerate 48000 doesn't match target 44100, resampling with SoXR
2024-01-02 00:41:04,775 (separator.py) (Line 447) [INFO] : Running separation...

digression

Whether the GUI has a multilingual support plan

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 1, 2024 •
edited
Loading

Whether the GUI has a multilingual support plan

It would be a lot of work to do so, as currently all the texts are hard-coded, but it is doable

The library of intel-extension-for-pytorch is not quite stable currently, actually

@DDXDB DDXDB changed the title MKL版本不稳定 MKL Version instability Jan 2, 2024
@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 4, 2024

@DDXDB Please try the build uploaded just now.

@DDXDB
Copy link
Author

DDXDB commented Jan 5, 2024

No longer crashes, but seems too slow, 5 minutes is still 0%.
Still sometimes the GPU is not recognized.

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 5, 2024 •
edited
Loading

OK

seems too slow, 5 minutes is still 0%.

Please be patient while I'm writing documentation for this issue

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 5, 2024

In short, for A770, download 12.55.0_dg2-g10-a0.7z from https://www.fosshub.com/Demucs-GUI-old.html and extract it and replace the file in Demucs-GUI.dir\intel_extension_for_pytorch\bin

@DDXDB
Copy link
Author

DDXDB commented Jan 5, 2024

Correction: After a long 0%, 8 minutes of audio was completed in about 30 seconds

@DDXDB
Copy link
Author

DDXDB commented Jan 5, 2024 •
edited
Loading

Uses the original Releases still, no replacement files.
13 minutes 40 seconds of video, holding 0% after about 6 minutes the GPU starts working, and it finishes in 30 seconds

2024-01-05 23:53:12,950 (GuiMain.py) (Line 1412) [INFO] : Python version: 3.11.7 | packaged by conda-forge | (main, Dec 23 2023, 14:27:59) [MSC v.1937 64 bit (AMD64)]
2024-01-05 23:53:12,950 (GuiMain.py) (Line 1413) [INFO] : Demucs GUI version: 1.1a2
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1414) [INFO] : System: Windows-10-10.0.22621-SP0
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1415) [INFO] : Architecture: 64bit
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1416) [INFO] : CPU: AMD64 Family 25 Model 33 Stepping 0, AuthenticAMD
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1417) [INFO] : CPU count: 12
2024-01-05 23:53:13,004 (GuiMain.py) (Line 1418) [INFO] : System memory: 34266697728 (31.913GB)
2024-01-05 23:53:13,008 (GuiMain.py) (Line 1421) [INFO] : System free memory: 20693766144 (19.273GB)
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1424) [INFO] : System swap memory: 38654705664 (36GB)
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1435) [INFO] : Using PySide6
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1436) [INFO] : Qt version: 6.6.1
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1437) [INFO] : PySide6 version: 6.6.1
2024-01-05 23:53:13,744 (shared.py) (Line 243) [INFO] : [1] Thread starter (separator.py) starts
2024-01-05 23:53:13,809 (GuiMain.py) (Line 1442) [DEBUG] : Supported styles: windowsvista, Windows, Fusion
2024-01-05 23:53:13,810 (GuiMain.py) (Line 1446) [DEBUG] : Current style: windowsvista
2024-01-05 23:53:18,281 (separator.py) (Line 53) [INFO] : Intel Extension for PyTorch version: 2.1.10+git45400a8
2024-01-05 23:53:18,370 (audio.py) (Line 34) [INFO] : Soundfile version: 0.12.1
2024-01-05 23:53:18,370 (audio.py) (Line 35) [INFO] : libsndfile version: 1.2.0
2024-01-05 23:53:18,370 (audio.py) (Line 36) [INFO] : SoXR version: 0.3.7
2024-01-05 23:53:18,370 (audio.py) (Line 37) [INFO] : libsoxr version: 0.1.3-7-gacc0dac
2024-01-05 23:53:18,370 (separator.py) (Line 63) [INFO] : Demucs version: 4.1.0a2
2024-01-05 23:53:18,371 (separator.py) (Line 64) [INFO] : PyTorch version: 2.1.0a0+cxx11.abi
2024-01-05 23:53:18,371 (separator.py) (Line 79) [INFO] : Intel MKL Info: _DeviceProperties(name='Intel(R) Arc(TM) A770 Graphics', platform_name='Intel(R) Level-Zero', dev_type='gpu, support_fp64=0, total_memory=15930MB, max_compute_units=512, gpu_eu_count=512)
2024-01-05 23:53:19,413 (audio.py) (Line 49) [INFO] : ffmpeg -version output:
ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil      58.  2.100 / 58.  2.100
libavcodec     60.  3.100 / 60.  3.100
libavformat    60.  3.100 / 60.  3.100
libavdevice    60.  1.100 / 60.  1.100
libavfilter     9.  3.100 /  9.  3.100
libswscale      7.  1.100 /  7.  1.100
libswresample   4. 10.100 /  4. 10.100
libpostproc    57.  1.100 / 57.  1.100

2024-01-05 23:53:19,456 (audio.py) (Line 54) [INFO] : Using ffmpeg from D:\Program Files\Demucs-GUI_1.1a2_mkl\ffmpeg\ffmpeg.EXE
2024-01-05 23:53:19,456 (audio.py) (Line 55) [INFO] : ffprobe -version output:
ffprobe version 6.0-full_build-www.gyan.dev Copyright (c) 2007-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil      58.  2.100 / 58.  2.100
libavcodec     60.  3.100 / 60.  3.100
libavformat    60.  3.100 / 60.  3.100
libavdevice    60.  1.100 / 60.  1.100
libavfilter     9.  3.100 /  9.  3.100
libswscale      7.  1.100 /  7.  1.100
libswresample   4. 10.100 /  4. 10.100
libpostproc    57.  1.100 / 57.  1.100

2024-01-05 23:53:20,457 (shared.py) (Line 254) [INFO] : [1] Thread starter (separator.py) ends
2024-01-05 23:53:20,460 (shared.py) (Line 243) [INFO] : [2] Thread checkUpdate (shared.py) starts
2024-01-05 23:53:20,472 (shared.py) (Line 270) [INFO] : Checking for updates...
2024-01-05 23:53:22,464 (shared.py) (Line 276) [WARNING] : Failed to check for updates:
Traceback (most recent call last):
  File "shared.py", line 271, in checkUpdate
  File "urllib\request.py", line 216, in urlopen
  File "urllib\request.py", line 525, in open
  File "urllib\request.py", line 634, in http_response
  File "urllib\request.py", line 563, in error
  File "urllib\request.py", line 496, in _call_chain
  File "urllib\request.py", line 643, in http_error_default
urllib.error.HTTPError: HTTP Error 403: rate limit exceeded

2024-01-05 23:53:22,465 (shared.py) (Line 254) [INFO] : [2] Thread checkUpdate (shared.py) ends
2024-01-05 23:54:33,095 (shared.py) (Line 243) [INFO] : [3] Thread loadModel (GuiMain.py) starts
2024-01-05 23:54:33,097 (GuiMain.py) (Line 505) [INFO] : Loading model htdemucs from repo D:\Program Files\Demucs-GUI_1.1a2_mkl\pretrained
2024-01-05 23:54:33,460 (GuiMain.py) (Line 527) [INFO] : Model loaded within 0.3626s
2024-01-05 23:54:33,460 (GuiMain.py) (Line 528) [INFO] : Model: htdemucs
Repo: D:\Program Files\Demucs-GUI_1.1a2_mkl\pretrained
Type: Bag of models
Audio channels: 2
Sample rate: 44100
Sources: drums, bass, other, vocals

Model 0:
　Type: HTDemucs
　Default segment: 7.8
　Weight: [1.0, 1.0, 1.0, 1.0]
2024-01-05 23:54:33,460 (shared.py) (Line 254) [INFO] : [3] Thread loadModel (GuiMain.py) ends
2024-01-05 23:56:07,428 (shared.py) (Line 243) [INFO] : [4] Thread separate (separator.py) starts
2024-01-05 23:56:07,428 (separator.py) (Line 415) [INFO] : Start separating audio: 1.mp3
2024-01-05 23:56:07,428 (separator.py) (Line 416) [INFO] : Parameters: segment=7.80 overlap=0.25 shifts=0
2024-01-05 23:56:07,428 (separator.py) (Line 417) [INFO] : Device: xpu:0
2024-01-05 23:56:07,428 (audio.py) (Line 66) [DEBUG] : Reading audio with soundfile: D:\1.mp3
2024-01-05 23:56:08,003 (audio.py) (Line 82) [INFO] : Read audio D:\1.mp3: samplerate=48000 shape=(39393280, 2)
2024-01-05 23:56:08,003 (audio.py) (Line 85) [INFO] : Samplerate 48000 doesn't match target 44100, resampling with SoXR
2024-01-05 23:56:09,602 (separator.py) (Line 447) [INFO] : Running separation...
2024-01-06 00:02:04,962 (separator.py) (Line 472) [INFO] : Saving separated audio...
2024-01-06 00:02:04,963 (shared.py) (Line 243) [INFO] : [5] Thread save (GuiMain.py) starts
2024-01-06 00:02:04,968 (shared.py) (Line 254) [INFO] : [4] Thread separate (separator.py) ends
2024-01-06 00:02:04,968 (GuiMain.py) (Line 1221) [INFO] : Mixing stem vocals
2024-01-06 00:02:07,731 (audio.py) (Line 129) [INFO] : Saved audio 2024-01-05 23:53:12,950 (GuiMain.py) (Line 1412) [INFO] : Python version: 3.11.7 | packaged by conda-forge | (main, Dec 23 2023, 14:27:59) [MSC v.1937 64 bit (AMD64)]
2024-01-05 23:53:12,950 (GuiMain.py) (Line 1413) [INFO] : Demucs GUI version: 1.1a2
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1414) [INFO] : System: Windows-10-10.0.22621-SP0
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1415) [INFO] : Architecture: 64bit
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1416) [INFO] : CPU: AMD64 Family 25 Model 33 Stepping 0, AuthenticAMD
2024-01-05 23:53:12,997 (GuiMain.py) (Line 1417) [INFO] : CPU count: 12
2024-01-05 23:53:13,004 (GuiMain.py) (Line 1418) [INFO] : System memory: 34266697728 (31.913GB)
2024-01-05 23:53:13,008 (GuiMain.py) (Line 1421) [INFO] : System free memory: 20693766144 (19.273GB)
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1424) [INFO] : System swap memory: 38654705664 (36GB)
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1435) [INFO] : Using PySide6
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1436) [INFO] : Qt version: 6.6.1
2024-01-05 23:53:13,428 (GuiMain.py) (Line 1437) [INFO] : PySide6 version: 6.6.1
2024-01-05 23:53:13,744 (shared.py) (Line 243) [INFO] : [1] Thread starter (separator.py) starts
2024-01-05 23:53:13,809 (GuiMain.py) (Line 1442) [DEBUG] : Supported styles: windowsvista, Windows, Fusion
2024-01-05 23:53:13,810 (GuiMain.py) (Line 1446) [DEBUG] : Current style: windowsvista
2024-01-05 23:53:18,281 (separator.py) (Line 53) [INFO] : Intel Extension for PyTorch version: 2.1.10+git45400a8
2024-01-05 23:53:18,370 (audio.py) (Line 34) [INFO] : Soundfile version: 0.12.1
2024-01-05 23:53:18,370 (audio.py) (Line 35) [INFO] : libsndfile version: 1.2.0
2024-01-05 23:53:18,370 (audio.py) (Line 36) [INFO] : SoXR version: 0.3.7
2024-01-05 23:53:18,370 (audio.py) (Line 37) [INFO] : libsoxr version: 0.1.3-7-gacc0dac
2024-01-05 23:53:18,370 (separator.py) (Line 63) [INFO] : Demucs version: 4.1.0a2
2024-01-05 23:53:18,371 (separator.py) (Line 64) [INFO] : PyTorch version: 2.1.0a0+cxx11.abi
2024-01-05 23:53:18,371 (separator.py) (Line 79) [INFO] : Intel MKL Info: _DeviceProperties(name='Intel(R) Arc(TM) A770 Graphics', platform_name='Intel(R) Level-Zero', dev_type='gpu, support_fp64=0, total_memory=15930MB, max_compute_units=512, gpu_eu_count=512)
2024-01-05 23:53:19,413 (audio.py) (Line 49) [INFO] : ffmpeg -version output:
ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil      58.  2.100 / 58.  2.100
libavcodec     60.  3.100 / 60.  3.100
libavformat    60.  3.100 / 60.  3.100
libavdevice    60.  1.100 / 60.  1.100
libavfilter     9.  3.100 /  9.  3.100
libswscale      7.  1.100 /  7.  1.100
libswresample   4. 10.100 /  4. 10.100
libpostproc    57.  1.100 / 57.  1.100

2024-01-05 23:53:19,456 (audio.py) (Line 54) [INFO] : Using ffmpeg from D:\Program Files\Demucs-GUI_1.1a2_mkl\ffmpeg\ffmpeg.EXE
2024-01-05 23:53:19,456 (audio.py) (Line 55) [INFO] : ffprobe -version output:
ffprobe version 6.0-full_build-www.gyan.dev Copyright (c) 2007-2023 the FFmpeg developers
built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
libavutil      58.  2.100 / 58.  2.100
libavcodec     60.  3.100 / 60.  3.100
libavformat    60.  3.100 / 60.  3.100
libavdevice    60.  1.100 / 60.  1.100
libavfilter     9.  3.100 /  9.  3.100
libswscale      7.  1.100 /  7.  1.100
libswresample   4. 10.100 /  4. 10.100
libpostproc    57.  1.100 / 57.  1.100

2024-01-05 23:53:20,457 (shared.py) (Line 254) [INFO] : [1] Thread starter (separator.py) ends
2024-01-05 23:53:20,460 (shared.py) (Line 243) [INFO] : [2] Thread checkUpdate (shared.py) starts
2024-01-05 23:53:20,472 (shared.py) (Line 270) [INFO] : Checking for updates...
2024-01-05 23:53:22,464 (shared.py) (Line 276) [WARNING] : Failed to check for updates:
Traceback (most recent call last):
  File "shared.py", line 271, in checkUpdate
  File "urllib\request.py", line 216, in urlopen
  File "urllib\request.py", line 525, in open
  File "urllib\request.py", line 634, in http_response
  File "urllib\request.py", line 563, in error
  File "urllib\request.py", line 496, in _call_chain
  File "urllib\request.py", line 643, in http_error_default
urllib.error.HTTPError: HTTP Error 403: rate limit exceeded

2024-01-05 23:53:22,465 (shared.py) (Line 254) [INFO] : [2] Thread checkUpdate (shared.py) ends
2024-01-05 23:54:33,095 (shared.py) (Line 243) [INFO] : [3] Thread loadModel (GuiMain.py) starts
2024-01-05 23:54:33,097 (GuiMain.py) (Line 505) [INFO] : Loading model htdemucs from repo D:\Program Files\Demucs-GUI_1.1a2_mkl\pretrained
2024-01-05 23:54:33,460 (GuiMain.py) (Line 527) [INFO] : Model loaded within 0.3626s
2024-01-05 23:54:33,460 (GuiMain.py) (Line 528) [INFO] : Model: htdemucs
Repo: D:\Program Files\Demucs-GUI_1.1a2_mkl\pretrained
Type: Bag of models
Audio channels: 2
Sample rate: 44100
Sources: drums, bass, other, vocals

Model 0:
　Type: HTDemucs
　Default segment: 7.8
　Weight: [1.0, 1.0, 1.0, 1.0]
2024-01-05 23:54:33,460 (shared.py) (Line 254) [INFO] : [3] Thread loadModel (GuiMain.py) ends
2024-01-05 23:56:07,428 (shared.py) (Line 243) [INFO] : [4] Thread separate (separator.py) starts
2024-01-05 23:56:07,428 (separator.py) (Line 415) [INFO] : Start separating audio: 1.mp3
2024-01-05 23:56:07,428 (separator.py) (Line 416) [INFO] : Parameters: segment=7.80 overlap=0.25 shifts=0
2024-01-05 23:56:07,428 (separator.py) (Line 417) [INFO] : Device: xpu:0
2024-01-05 23:56:07,428 (audio.py) (Line 66) [DEBUG] : Reading audio with soundfile: D:\1.mp3
2024-01-05 23:56:08,003 (audio.py) (Line 82) [INFO] : Read audio D:\1.mp3: samplerate=48000 shape=(39393280, 2)
2024-01-05 23:56:08,003 (audio.py) (Line 85) [INFO] : Samplerate 48000 doesn't match target 44100, resampling with SoXR
2024-01-05 23:56:09,602 (separator.py) (Line 447) [INFO] : Running separation...
2024-01-06 00:02:04,962 (separator.py) (Line 472) [INFO] : Saving separated audio...
2024-01-06 00:02:04,963 (shared.py) (Line 243) [INFO] : [5] Thread save (GuiMain.py) starts
2024-01-06 00:02:04,968 (shared.py) (Line 254) [INFO] : [4] Thread separate (separator.py) ends
2024-01-06 00:02:04,968 (GuiMain.py) (Line 1221) [INFO] : Mixing stem vocals
2024-01-06 00:02:07,731 (audio.py) (Line 129) [INFO] : Saved audio D:\separated\htdemucs\1\vocals.flac: shape=torch.Size([2, 36192576])
2024-01-06 00:02:07,731 (shared.py) (Line 254) [INFO] : [5] Thread save (GuiMain.py) ends
1\vocals.flac: shape=torch.Size([2, 36192576])
2024-01-06 00:02:07,731 (shared.py) (Line 254) [INFO] : [5] Thread save (GuiMain.py) ends

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 5, 2024 •
edited
Loading

Yes, it's due to the AOT issue. When the progress gets stuck at 0%, torch (IPEX, precisely) is compiling the acceleration codes. The replacement contains the compiled codes so it won't need to compile it before running separation anymore :)
Besides, I'd also like to ask you to try replacing the file and see whether it get stuck anymore

@DDXDB
Copy link
Author

DDXDB commented Jan 5, 2024

Yes, it's due to the AOT issue. When the progress gets stuck at 0%, torch (IPEX, precisely) is compiling the acceleration codes. The replacement contains the compiled codes so it won't need to compile it before running separation anymore :) Besides, I'd also like to ask you to try replacing the file and see whether it get stuck anymore

Same result

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 5, 2024 •
edited
Loading

What about 12.55.8?
12.55.0 is what the compiler outputs when entering the PCI ID of A770 (0x56A0), but according to the architecture of A770 it should be 12.55.8

@DDXDB
Copy link
Author

DDXDB commented Jan 5, 2024

What about 12.55.8? 12.55.0 is what the compiler outputs when entering the PCI ID of A770 (0x56A0), but according to the architecture of A770 it should be 12.55.8

Work began almost immediately and took a slightly longer time, taking 50 seconds to complete

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 6, 2024

Thanks!
I understand. Here is my deduction: Local JIT can make sure that the compiled codes are optimized for a specific GPU, which runs more efficiently, but takes a long time to prepare. Using AOT can save time on building the codes, but is not as efficient as JIT. Just like Gentoo Linux, compiling everything locally to increase performance.

@DDXDB
Copy link
Author

DDXDB commented Jan 6, 2024

Thanks! I understand. Here is my deduction: Local JIT can make sure that the compiled codes are optimized for a specific GPU, which runs more efficiently, but takes a long time to prepare. Using AOT can save time on building the codes, but is not as efficient as JIT. Just like Gentoo Linux, compiling everything locally to increase performance.

Can I save it after compiling locally? You need to recompile the application every time you restart it.

@CarlGao4
Copy link
Owner

CarlGao4 commented Jan 6, 2024

Unfortunately, I'm also trying to find out how Intel's JIT works.

@CarlGao4
Copy link
Owner

CarlGao4 commented Sep 5, 2024

Found solution. Will start developing within days

@CarlGao4
Copy link
Owner

CarlGao4 commented Nov 2, 2024

Bad news. There is a bug inside the OpenCL compiler redistributed along with the Intel graphics driver on Windows, which makes it impossible to reuse the cache. I've submitted it to Intel official and they have fixed it, but not sure when it will be released to consumers so currently the new codes are still on my local disk and haven't been pushed to GitHub. As Windows requires all its drivers should be WHQL certificated, waiting for official driver update should be the only option.

@CarlGao4 CarlGao4 mentioned this issue Jan 10, 2025
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
bug Something isn't working
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@DDXDB @CarlGao4