README: retally the numbers after the dispatch and render work - #6
Merged
Conversation
Re-measured against the pinned upstream on a fullscreen canvas with the merged optimizations in. The median across the 35 work-bound effects moves from 14.1x to 27.5x (range 17.1x-47.4x), and the four effects in the table roughly halve again: rings 1,492 ms -> 521 ms, waves 1,082 -> 374, beams 345 -> 181, slide 147 -> 76. Also corrects what the clock-bound line claims. matrix and thunderstorm report 1.9x and 1.3x, but that is the harness's wall-clock ratio, not a count of frames rendered — the harness never counts Python's frames. Most of their runtime is a fixed animation duration neither side can shorten, so the sentence now says that instead. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Re-measured against the pinned upstream reference on a fullscreen canvas with
PR #5 merged, using the repo's own harness:
One correction beyond the numbers
The clock-bound sentence said
matrixandthunderstorm"render 2.1× and 1.2×the frames in the same seconds." That isn't what the harness measures — it times
Python but never counts Python's frames, so the ratio in that column is wall
clock, and the two sides don't run for equal durations anyway (matrix: 15.1 s
vs 28.9 s). I rewrote the sentence to say what the number is: most of their
runtime is a fixed animation duration neither implementation can shorten, which
is why they come in at 1.9× and 1.3×.
I did try to measure the frame counts directly, by counting the per-frame
DEC-restore-cursor in each side's tty stream. It works, but with pacing off the
reader becomes the bottleneck, so the counts measure the pipe rather than the
renderer — and the ttfx run filled a 31 GB tmpfs on the way. Not a number worth
publishing, so the README no longer claims one.
— 🤖 Claude, posting on behalf of @dhh