Page 1 of 1

During video conversion, the system memory is consumed little by little....

Posted: Mon Jun 22, 2020 11:46 am
by MerrisonCat

During the conversion process, memory resources are consumed a little. It is expected that 10 minutes after the execution starts, memory will be exhausted and errors will occur!

The video is about 20 minutes long and 800MB in size.

With Windows10, it takes about 2 minutes to burn memory; with Ubuntu, it takes about 10 minutes...
Please ask me how to solve this problem...

============ System Information ============
encoding: UTF-8
git_branch: master
git_commits: 127d3db Dependencies update (#1028). ac40b0f Remove subpixel upscaling option (#1024). 92bc9af Extraction: Added an option to skip outputting face images (#1021). 815c843 Simple backend unit tests (#1020). d27897a Slight tweak of wording
gpu_cuda: 10.0
gpu_cudnn: 7.6.4
gpu_devices: GPU_0: GeForce GTX TITAN X
gpu_devices_active: GPU_0
gpu_driver: 440.82
gpu_vram: GPU_0: 12211MB
os_machine: x86_64
os_platform: Linux-5.3.0-59-generic-x86_64-with-debian-buster-sid
os_release: 5.3.0-59-generic
py_command: /home/pc-gmxy/ShareFileOS/faceswap/faceswap.py convert -i /home/pc-gmxy/ShareFileOS/Source/WeiJiu_Src/Vedio.mp4 -o /home/pc-gmxy/ShareFileOS/Source/FramOutput -m /home/pc-gmxy/ShareFileOS/Model -c none -M none -sc sharpen -w ffmpeg -osc 100 -l 0.2 -j 0 -g 1 -t original -otf -L INFO -gui
py_conda_version: conda 4.8.2
py_implementation: CPython
py_version: 3.7.7
py_virtual_env: True
sys_cores: 8
sys_processor: x86_64
sys_ram: Total: 23994MB, Available: 3429MB, Used: 19024MB, Free: 810MB


Re: During video conversion, the system memory is consumed little by little....

Posted: Mon Jun 22, 2020 4:48 pm
by bryanlyon

This is normal as Python dynamically allocates memory with garbage collection occurring when needed. You wont actually run out of memory but conversion may slow down a bit as garbage is collected to clear the extra memory and it continues.


Re: During video conversion, the system memory is consumed little by little....

Posted: Thu Jun 25, 2020 7:11 am
by MerrisonCat

Finally, I tried to divide the video into multiple segments (4 segments), processed it and succeeded.
Although it is possible to use a smaller number of threads (2 threads), the processing speed is really unacceptable!