impr: Use f16 and better subgroup shader in MNIST example#2412
Open
reczkok wants to merge 3 commits into
Open
Conversation
|
pkg.pr.new packages benchmark commit |
📊 Bundle Size Comparison
👀 Notable resultsStatic test results:No major changes. Dynamic test results:No major changes. 📋 All resultsClick to reveal the results table (354 entries).
If you wish to run a comparison for other, slower bundlers, run the 'Tree-shake test' from the GitHub Actions menu. |
Resolution Time Benchmark---
config:
themeVariables:
xyChart:
plotColorPalette: "#E63946, #3B82F6, #059669"
---
xychart
title "Random Branching (🔴 PR | 🔵 main | 🟢 release)"
x-axis "max depth" [1, 2, 3, 4, 5, 6, 7, 8]
y-axis "time (ms)"
line [0.95, 1.93, 4.42, 6.49, 7.69, 11.01, 22.11, 22.52]
line [0.99, 2.06, 4.40, 6.69, 7.60, 11.03, 21.56, 24.80]
line [0.91, 1.89, 4.39, 6.48, 7.36, 10.73, 21.20, 22.31]
---
config:
themeVariables:
xyChart:
plotColorPalette: "#E63946, #3B82F6, #059669"
---
xychart
title "Linear Recursion (🔴 PR | 🔵 main | 🟢 release)"
x-axis "max depth" [1, 2, 3, 4, 5, 6, 7, 8]
y-axis "time (ms)"
line [0.34, 0.53, 0.64, 0.81, 1.13, 1.18, 1.40, 1.61]
line [0.29, 0.54, 0.71, 0.87, 1.18, 1.25, 1.47, 1.58]
line [0.31, 0.52, 0.71, 0.87, 1.12, 1.14, 1.38, 1.53]
---
config:
themeVariables:
xyChart:
plotColorPalette: "#E63946, #3B82F6, #059669"
---
xychart
title "Full Tree (🔴 PR | 🔵 main | 🟢 release)"
x-axis "max depth" [1, 2, 3, 4, 5, 6, 7, 8]
y-axis "time (ms)"
line [0.89, 2.07, 4.26, 6.69, 12.73, 26.66, 56.97, 119.53]
line [0.82, 2.17, 3.63, 6.59, 12.54, 26.94, 55.85, 114.15]
line [0.83, 2.14, 4.31, 6.51, 13.08, 27.06, 55.75, 113.60]
|
reczkok
force-pushed
the
impr/cleaner-mnist
branch
from
79b1935 to
4328747
Compare
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+40
to
+43
| export function downloadLayers( | ||
| root: TgpuRoot, | ||
| floatShcema: d.F32 | d.F16, | ||
| ): Promise<[LayerData, LayerData][]> { |
There was a problem hiding this comment.
Typo in parameter name floatShcema (should be floatSchema). Keeping the misspelling makes the API harder to read/search and increases the chance of propagating the typo to call sites.
Comment on lines
+179
to
+182
| const outputCount = buffers[i].biases.dataType.elementCount; | ||
| boundPipeline.dispatchWorkgroups( | ||
| subgroupPipeline ? outputCount : Math.ceil(outputCount / WORKGROUP_SIZE), | ||
| ); |
There was a problem hiding this comment.
dispatchWorkgroups uses outputCount when the subgroup pipeline is selected, but subgroupCompute computes num_subgroups outputs per workgroup (neuronIndex = wid.x * nsg + sgid). This over-dispatches workgroups by a factor of nsg (e.g., 2x for 64 threads with 32-wide subgroups), doing unnecessary work for larger layers. Consider either dispatching ceil(outputCount / outputsPerWorkgroup) (if you can determine outputsPerWorkgroup) or adjusting the shader/work mapping so each workgroup corresponds to exactly one output when dispatch count must be outputCount.
aleksanderkatan
approved these changes
May 14, 2026
| const context = canvas.getContext('2d') as CanvasRenderingContext2D; | ||
|
|
||
| const bars = Array.from(document.querySelectorAll('.bar')) as HTMLDivElement[]; | ||
| const subgroupsEl = document.getElementById('subgroups-status') as HTMLSpanElement; |
Contributor
There was a problem hiding this comment.
Could we also include status for f16? I don't think there is a way to know whether the shader runs on f16 or f32 at this moment
reczkok
force-pushed
the
impr/cleaner-mnist
branch
from
4328747 to
82d84e8
Compare
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
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.
No description provided.