Debug c++ arch performance bottlenecks

[mkopcins]

Description

Currently we are seeing big performance drops (around 50%) when running LLMs on v0.5 vs v0.4

Steps to reproduce

1. checkout main branch
2. run LLM app

Snack or a link to a repository

No response

React Native Executorch version

0.5.0

React Native version

0.73.1

Platforms

Platform agnostic

JavaScript runtime

None

Workflow

None

Architecture

None

Build type

None

Device

None

Device model

No response

AI model

No response

Performance logs

No response

Acknowledgements

Yes