docs: Add guide for profiling MLX models with Xcode Instruments #1395
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This guide demonstrates how to use Metal System Trace to identify CPU dispatch overhead vs fused kernels, helping developers optimize MLX models on Apple Silicon. Includes: - Step-by-step profiling workflow using xctrace CLI - Benchmark script demonstrating kernel fusion benefits - Visual comparison of eager vs compiled execution patterns
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation
While analyzing MLX performance on my Mac Studio (M4 Max), I realized that visualizing GPU execution patterns is critical for understanding optimization. Currently, there seems to be a lack of documentation on how to leverage Xcode Instruments with MLX.
Changes
guides/profiling_with_instruments.mdContext
I am a student aiming to become an inference optimization engineer. I found that MLX's kernel fusion drastically reduces memory bandwidth pressure compared to PyTorch on Unified Memory architectures. I hope this guide helps other developers optimize their models.
Thank you for your hard work on this amazing framework!