Skip to content

Reduce per-frame allocations and release size - #4

Merged
RamesTheGeneric merged 1 commit into
Project-Babble:mainfrom
cinnacappu:optimize-runtime
Oct 21, 2026
Merged

RamesTheGeneric merged 1 commit into
Project-Babble:mainfrom
cinnacappu:optimize-runtime

Conversation

@cinnacappu

Copy link
Copy Markdown
Contributor

Reuse snapshots and polling buffers, move inference arguments onto the stack, and allocate mouth-processing scratch space only when needed. Also avoid repeated resize and image-blending work. Model inputs, filtering, calibration and packet formats stay unchanged.

Strip unused symbol tables from Release binaries while keeping dynamic exports and crash backtrace names. Applying the same stripping to the current release saved about 2.54 MB installed and 227 KB compressed; the rebuilt package may differ.

Local checks matched the original preprocessing results, OSC packets and dashboard pixels. Concurrent snapshot checks and existing image/dashboard tests passed, and all changed C++ files compile for Linux ARM64. A full linked build and testing on the Frame are still pending. No test or workflow changes are included.

@RamesTheGeneric

Copy link
Copy Markdown
Contributor

Hey, just letting you know I saw this PR. I should get a chance to review this later today.

@RamesTheGeneric
RamesTheGeneric merged commit 8a36091 into Project-Babble:main Oct 21, 2026
@cinnacappu
cinnacappu deleted the optimize-runtime branch October 21, 2026 04:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants