stable-diffusion VS tvm

SD on an Intel mac with Vega graphics runs pretty well though — I think it ran at something like ~3-5 iterations/s for me, which is decent. I ran either https://github.com/magnusviri/stable-diffusion or https://github.com/lstein/stable-diffusion which have MPS support

https://github.com/magnusviri/stable-diffusion/commit/d0b168...

Copying this change fixed seeds on M1 for me.

You should be able to run it on a CPU. Maybe try this version. If MPS is supported on your Mac you can check this out.

A group of open source hackers forked Stable Diffusion on GitHub and optimized the model to run on Apple's M1 chip, enabling images to be generated in ~ 15 seconds (512x512 pixels, 50 diffusion steps).

Magnusviro [0], the original author of the SD M1 repo credited in this article, has merged his fork into the Lstein Stable Diffusion repo [1], and you can now run Lstein fork with M1 as of a few hours ago.

This adds a ton of functionality - GUI, Upscaling & Facial improvements, weighted subprompts etc.

This has been a big undertaking over the last few days, and I highly recommend checking it out.

[0] https://github.com/magnusviri/stable-diffusion

You can run it on an M1. Using a macbook M1 pro max with 32Gb I get 512x512 in about 50 seconds. use this branch https://github.com/magnusviri/stable-diffusion/tree/apple-mps-support

I had this error too, and I tried a ton of things to get cudatoolkit to install, without any luck. This fork has an environment-mac.yml file that actually got it working on my M1 Max: https://github.com/magnusviri/stable-diffusion/tree/apple-silicon-mps-support

But when I run it (locally, using the Mac M1 port), every time I run it creates a different image.

Yes. Web-llm is a wrapper of tvmjs: https://github.com/apache/tvm

Just wrappers all the way down

Yes, this is coming! Myself and others at OctoML and in the TVM community are actively working on multi-gpu support in the compiler and runtime. Here are some of the merged and active PRs on the multi-GPU (multi-device) roadmap:

Support in TVM’s graph IR (Relax) - https://github.com/apache/tvm/pull/15447

Would it make sense to have a backend support for OpenXLA, Apache TVM, Jittor or other similar to get free GPU, TPU and other accelerators for free ?

I have tried the iPhone app. It's fast. They're using Apache TVM which should allow better use of native accelerators on different devices. Like using metal on Apple and Vulcan or CUDA or whatever instead of just running the thing on the CPU like llama.cpp.

I was going to answer the same, I find the approach of machine learning compilers that directly compile models to host and device code better than having to bring a huge runtime. There are exciting projects in this area like TVM Unity, IREE [2], or torch.export [3]

[1] https://github.com/apache/tvm/tree/unity

[2] https://pytorch.org/get-started/pytorch-2.0/#inference-and-e...

[3] https://pytorch.org/get-started/pytorch-2.0/#inference-and-e...

Apache TVM home page: https://tvm.apache.org/

It's pretty clear its referring to the output of Apache TVM and Meta's Glow

As mentioned in sibling comments, Torch is indeed the glue in this implementation. Other glues are TVM[0] and ONNX[1]

These just cover the neural net though, and there is lots of surrounding code and pre-/post-processing that isn't covered by these systems.

For models on Replicate, we use Docker, packaged with Cog for this stuff.[2] Unfortunately Docker doesn't run natively on Mac, so if we want to use the Mac's GPU, we can't use Docker.

I wish there was a good container system for Mac. Even better if it were something that spanned both Mac and Linux. (Not as far-fetched as it seems... I used to work at Docker and spent a bit of time looking into this...)

[0] https://tvm.apache.org/

Or use TVM, the idea is to compile your model into code that you can load at runtime. Similar to onnxruntime, it only does DNN inference; so you need domain-specific code.