Repository navigation
Conversation
…e update-loop benchmark The instanced path is bound by the JavaScript update loop. The handoff records where the frame goes, a candidate that removes 27% of the update time with identical output, and the opcode profile to take to GocciaScript. benchmarks/bunnymark-update-loop.ts reproduces the measurement without a window or raylib.
|
Navigate logical layers of code changes, visualize relationships, and explore their blast radius. Note Currently processing new changes in this PR. This may take a few minutes, please wait... ⚙️ Run configurationConfiguration used: Central YAML (base), Organization UI (inherited) Review profile: CHILL Plan: Team Run ID: 📒 Files selected for processing (2)
Comment |
frostney
marked this pull request as ready for review
October 1, 2026 15:20
Owner
Author
|
Closing: this handoff should be a standalone markdown file, not a repository change. |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Bunnymark's instanced path is bound by the JavaScript update loop, not by rendering: on a Radeon RX 6600M, 27.6 ms of a 28.2 ms frame at 10,000 sprites is
updateBunnies. This records what was measured and what to do about it, so the optimisation can start in a fresh session. No example or binding code changes..agent/HANDOFF.mdis rewritten as the handoff for that work: where the frame goes, the change to make in this repository, the opcode profile to take to GocciaScript, and what was not verified.benchmarks/bunnymark-update-loop.tstimes the update loop alone, with no window, raylib or capability, and compares the current loop with a candidate that must produce identical state.Ryzen 9 6900HX, GocciaScript 0.14.0, 10,000 sprites, three rounds of 100 frames:
updateBunniesThe candidate is 27% faster with identical transforms, positions and velocities. It is not applied here; applying it and re-measuring across machines is the first step in the handoff.
The engine-side finding with the clearest evidence: a module-level
constread inside a function compiles toOP_GET_GLOBAL, which costs about 125 ns here against about 20 ns for an arithmetic opcode.