Mathieu Poumeyrol

I agree on principle. But I'm better at finding reason against documenting than actually documenting. :) Here, I could argue that the model costing is still experimental, that I may...

Running monocore is more or less by design. tract executor is monocore anyway, so it's closer to the "production" settings like that. I also hope it may help with overheating...

About the mmm/mmv... mmv is actually a harder problem to optimise, as most cpu and soc vendors will target the mmm use case: square-ish product have a higher multiplication/memory access...

DeepFilterNet2 is very challenging if you have a low-latency use case and need to run frame per frame. Grouping frames together (like pulse of 4 or 8) will improve performance...

These findings are interesting... I never spent too much time on the tiny-m, tiny-n case, but maybe it's time :) I think what we are observing here is memory access...

Hello, thanks for your interest in tract. What are we trying to achieve here ? If we want to integrate ONNX metadata with tract, there are a couple issue I...

OK, so this is not what I had in mind. What I was thinking was: 1/ load the ONNX metadata in parse_with_template, refactoring the operator set version extraction which is...

Could you give a shot at the current main branch ? I think this is solved.

Hello ! As far as I can tell this Reshape is invalid. 1,197,576 can not be reshaped as 1,197,3,3,65 : 3*3*65 = 585. The 3,3,65 shape is hard-coded as a...

Hello ! QLinearAdd is not an ONNX operator. See https://onnx.ai/onnx/operators/ for the list. Microsoft ONNXRuntime implements it as an extension, and there is an ongoing request to bring it to...