Sam Green green-s

Ubuntu 21.04 VM with GPU acceleration under Hyper-V...?

Modern versions of Windows support GPU paravirtualization in Hyper-V with normal consumer graphics cards. This is used e.g. for graphics acceleration in Windows Sandbox, as well as WSLg. In some cases, it may be useful to create a normal VM with GPU acceleration using this feature, but this is not officially supported. People already figured out how to do it with Windows guests though, so why not do the same with Linux? It should be easy given that WSLg is open source and reasonably well documented, right?

Well... not quite. I managed to get it to run... but not well.

How to do it?

Verify driver support

Stable Diffusion for AMD GPUs on Windows using DirectML

UPDATE: A faster (20x) approach for running Stable Diffusion using MLIR/Vulkan/IREE is available on Windows:

https://github.com/nod-ai/SHARK/blob/main/shark/examples/shark_inference/stable_diffusion/stable_diffusion_amd.md

Install 🤗 diffusers

conda create --name sd39 python=3.9 -y

Requirement

Console / Cloud Shell access (via https://cloud.oracle.com)
Go to the instance page and under Resources -> Console connection -> Launch Cloud Shell connection

Steps

In Ubuntu OR any other Free tier Linux OS

# Download Alpine Linux and install it on disk
cd /
wget https://dl-cdn.alpinelinux.org/alpine/v3.16/releases/x86_64/alpine-virt-3.16.2-x86_64.iso

	location /btsync/ {
	proxy_pass http://127.0.0.1:8888/gui/;
	sub_filter '/gui/token.html' 'token.html';
	sub_filter '/gui/' '';
	sub_filter_types text/javascript
	sub_filter_once off;
	proxy_set_header Cookie $http_cookie;
	proxy_set_header Host $host;
	proxy_set_header X-Real-IP $remote_addr;
	proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;

	# Attempts to demangle all mangled symbols in the current program using the Rust
	# mangling schemes, and replace the default symbol and function signature
	# (if applicable) with the demangled symbol.
	#
	# License: MIT OR Apache-2.0
	#@author Jack Grigg <thestr4d@gmail.com>
	#@category Symbol

	import string

	/*=============================================================================

	ReShade 5 effect file
	github.com/martymcmodding

	Author: Pascal Gilcher / Marty McFly

	ReShade Motion Estimation Shader for dense 2D UV-space motion vectors
	Based on ReShade Motion Estimation by Jakob Wapenhensch
	(https://github.com/JakobPCoder/ReshadeMotionEstimation)

	"""
	stable diffusion dreaming
	creates hypnotic moving videos by smoothly walking randomly through the sample space

	example way to run this script:

	$ python stablediffusionwalk.py --prompt "blueberry spaghetti" --name blueberry

	to stitch together the images, e.g.:
	$ ffmpeg -r 10 -f image2 -s 512x512 -i blueberry/frame%06d.jpg -vcodec libx264 -crf 10 -pix_fmt yuv420p blueberry.mp4

	def normalize_latent(x, max_val, quantile_val):
	x = x.detach().clone()
	for i in range(x.shape[0]):
	if x[[i], :].std() > 1.0:
	x[[i], :] = x[[i], :] / x[[i], :].std()
	s = torch.quantile(torch.abs(x[[i], :]), quantile_val)
	s = torch.maximum(s, torch.ones_like(s) * max_val)
	x[[i], :] = x[[i], :] / (s / max_val)
	return x

	# ffmpeg -i foo.m4a foo.wav

	from pyannote.audio import Pipeline

	pipeline = Pipeline.from_pretrained("pyannote/speaker-diarization")
	diarization = pipeline("foo.wav")

	# RTTM format https://catalog.ldc.upenn.edu/docs/LDC2004T12/RTTM-format-v13.pdf
	with open("foo.rttm", "w") as rttm:
	diarization.write_rttm(rttm)

	# Script for converting a HF Diffusers saved pipeline to a Stable Diffusion checkpoint.
	# Only converts the UNet, VAE, and Text Encoder.
	# Does not convert optimizer state or any other thing.
	# Written by jachiam

	import argparse
	import os.path as osp

	import torch