styler00dollar / VSGAN-tensorrt-docker

Using VapourSynth with super resolution and interpolation models and speeding them up with TensorRT.
BSD 3-Clause "New" or "Revised" License
288 stars 30 forks source link

Trying On A M1 Mac #14

Closed Ghostkwebb closed 2 years ago

Ghostkwebb commented 2 years ago

So I followed this tutorial https://www.youtube.com/watch?v=B134jvhO8yk&t=0s But when docker run --privileged --gpus all -it --rm -v /home/vsgan_path/:/workspace/tensorrt styler00dollar/vsgan_tensorrt:latest it just gives me an error that it doesn't find the right amd64 or somthing and I rage quit deleted it without seeing the full error. PLS HELP ME :(

Ghostkwebb commented 2 years ago

This is the error ghostkwebb@Sharads-MacBook-Pro VSGAN-tensorrt-docker % docker run --privileged --gpus all -it --rm -v //Users/ghostkwebb/Desktop/AI/VSGAN-tensorrt-docker:/workspace/tensorrt styler00dollar/vsgan_tensorrt:latest

WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested docker: Error response from daemon: could not select device driver "" with capabilities: [[gpu]].

pls help me I don't wanna pay money for fricking waifu2x

styler00dollar commented 2 years ago

My stuff was not made for arm in mind. And since amd gpus are not supported by docker, I kind of doubt that M1 gpus will work. The only chance you have is to manually compile the ncnn related vapoursynth packages on your system and use it without docker. Vapoursynth should be the most efficient way to use models, many other repos extract into images which is a waste of time and disk io bottleneck for read/write. TensorRT, docker and cuda and so on only works with nvidia.

Ghostkwebb commented 2 years ago

Could you tell me how to manually compile? Because there is the software waify2x which uses the stuff u made that's why I wanna know if I can do it too.

styler00dollar commented 2 years ago

https://github.com/Nlzy/vapoursynth-waifu2x-ncnn-vulkan

Ghostkwebb commented 2 years ago

I could install docker just fine but i don't know how to do the other things that your mentioned

On Sun, 28 Aug, 2022, 4:01 pm styler00dollar, @.***> wrote:

My stuff was not made for arm in mind. And since amd gpus are not supported by docker, I kind of doubt that M1 gpus will work. The only chance you have is to manually compile the ncnn related vapoursynth packages on your system and use it without docker. TensorRT and cuda and so on only works with nvidia.

— Reply to this email directly, view it on GitHub https://github.com/styler00dollar/VSGAN-tensorrt-docker/issues/14#issuecomment-1229428169, or unsubscribe https://github.com/notifications/unsubscribe-auth/AOVU72QSQKQ4GUWNNBB5IWDV3M5XTANCNFSM57ZT66WA . You are receiving this because you authored the thread.Message ID: @.***>

styler00dollar commented 2 years ago

The docker was not built for arm, so you can't use the docker and you need to do it manually. Either directly on your system or create a custom arm docker. I do not have an arm system and won't support it. Compile instructions are in the repository, but may differ for arm.

Docker does only support nvidia gpus though, so installing onto your system is most likely the way.