⚡ Profiler: Gaussian Splatting Loop Tiling #380
Conversation
- Implemented `blend_gaussians_block_2x2` in `rendering.rs` to process pixels in 2x2 blocks, reusing Gaussian data and reducing memory loads by 4x. - Added `#[inline]` to `evaluate_gaussian_opacity` and `blend_gaussians` for better cross-crate optimization. - Updated `bench_gaussian.rs` to benchmark the new tiled approach. - Added performance record to `.jules/profiler.md`. Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with New to Jules? Learn more at jules.google/docs. For security, I will only act on instructions from the user who triggered this task. |
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
📉 The Bottleneck: The
blend_gaussiansfunction was bandwidth-bound, reloading 88 bytes of Gaussian data for every single pixel (100M loads for 100 frames).🚀 The Boost: Before: 1.60s -> After: 1.56s (~2.7% improvement).
💻 Technical Detail: Implemented
blend_gaussians_block_2x2which processes 2x2 pixel blocks. This reuses the loaded Gaussian data (mean, conic, color) for 4 pixels, reducing memory loads by 4x. Also added#[inline]to hot math functions.🧪 Verification: Verified correctness with
cargo testand benchmark output comparison (results match). PDR entry added to.jules/profiler.md.PR created automatically by Jules for task 4031989690945417743 started by @fderuiter