Steward Garcia

It seems like your GPU doesn't support any of the matrix multiplication kernels, I'm not sure if it's due to lack of capabilities or compatibility with Metal.

That number is a small extra space for the data since some operations require padding; this is necessary when performing calculations with the context (without using ggml-alloc, which internally adds...

You're right, that 1024 should be the size of the output tensor data. Honestly, I'm not sure how to calculate it correctly before creating the context. @slaren Any idea on...

I think what should exist is documentation of the functions in ggml and their equivalents in PyTorch, and then each person can figure out how to port any model. The...

I think the metal backend could benefit from expanding matrix multiplication support and significantly improving performance in some cases such as [comment](https://github.com/leejet/stable-diffusion.cpp/issues/108#issuecomment-1853903260).

I could help you. First of all, the way you want to allocate memory to the context is wrong. ctx0 is for building the computation graph and only requires memory...

I'm going to follow up on this PR, although I'd prefer to at least understand the use case for this type of padding. According to @gartia , it's used in...

If this function proves to be useful and easy to integrate with the conv2d operation, it will be added. Since having an operation that no ggml project needs doesn't make...

Hello, may you help to use this library please, some example please. I want to contact with you.

Thank you, :), I hope it works