Skip to content

metal : add env var to trigger graph capture#20398

Merged
ggerganov merged 1 commit intomasterfrom
gg/metal-capture-env
Mar 11, 2026
Merged

metal : add env var to trigger graph capture#20398
ggerganov merged 1 commit intomasterfrom
gg/metal-capture-env

Conversation

@ggerganov
Copy link
Member

@ggerganov ggerganov commented Mar 11, 2026 •
edited
Loading

QoL for capturing execution of Metal graphs for profiling purposes.

Usage:

# capture the 8th compute graph
METAL_CAPTURE_ENABLED=1 GGML_METAL_CAPTURE_COMPUTE=8 llama-completion ...

0.00.707.463 W ggml_metal_graph_compute: capturing graph in /tmp/perf-metal-47193.gputrace

Inspect in XCode:

open /tmp/perf-metal-47193.gputrace

Useful for detecting register spills:

image

Or for finding hot spots in the kernel implementations:

image

@github-actions github-actions bot added ggml changes relating to the ggml tensor library for machine learning Apple Metal https://en.wikipedia.org/wiki/Metal_(API) labels Mar 11, 2026
@ggerganov ggerganov mentioned this pull request Mar 11, 2026
Merged
@ggerganov ggerganov merged commit c363256 into master Mar 11, 2026
16 of 75 checks passed
@ggerganov ggerganov deleted the gg/metal-capture-env branch March 11, 2026 14:25
This was referenced Mar 11, 2026
Merged
Merged
ProgenyAlpha pushed a commit to ProgenyAlpha/llama.cpp that referenced this pull request Mar 12, 2026
@ggerganov @ProgenyAlpha
metal : add env var to trigger graph capture (ggml-org#20398)
5c9e360
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@ggerganov