docs: Add an example of a fully GPU genetic algorithm#2250
Open
docs: Add an example of a fully GPU genetic algorithm#2250
Conversation
Contributor
reczkok
commented
Mar 5, 2026
reczkok
commented
|
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 (344 entries).
If you wish to run a comparison for other, slower bundlers, run the 'Tree-shake test' from the GitHub Actions menu. |
There was a problem hiding this comment.
Pull request overview
This PR adds a new interactive example demonstrating a fully GPU-driven genetic algorithm for 2D car racing simulation. Cars navigate procedurally generated tracks using sensor-based inputs processed through evolved genomes, with tournament selection, crossover, and mutation all running in GPU compute shaders.
Changes:
- Added a complete genetic racing example with procedural track generation (
track.ts), GPU-based genetic algorithm logic (ga.ts), and the main simulation/rendering loop (index.ts) - Included interactive controls for grid size, population, mutation parameters, and simulation speed, along with a car sprite asset and HTML overlay for stats display
- Added example metadata (
meta.json) and thumbnail for the docs example gallery
Reviewed changes
Copilot reviewed 5 out of 7 changed files in this pull request and generated no comments.
Show a summary per file
| File | Description |
|---|---|
apps/typegpu-docs/src/examples/algorithms/genetic-racing/meta.json |
Example metadata with title, category, and tags |
apps/typegpu-docs/src/examples/algorithms/genetic-racing/index.html |
Canvas element with stats overlay div and styling |
apps/typegpu-docs/src/examples/algorithms/genetic-racing/index.ts |
Main example: GPU pipeline setup, simulation loop, rendering, and user controls |
apps/typegpu-docs/src/examples/algorithms/genetic-racing/ga.ts |
Genetic algorithm: data schemas, init/evolve compute shaders, population management |
apps/typegpu-docs/src/examples/algorithms/genetic-racing/track.ts |
Procedural track generation with random walk paths and Catmull-Rom spline smoothing |
apps/typegpu-docs/public/assets/genetic-car/car.png |
Car sprite image asset |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
iwoplaza
changed the title
reczkok
force-pushed
the
docs/genetic-racing
branch
from
b235654 to
22abee6
Compare
reczkok
commented
Mar 7, 2026
Comment on lines
+408
to
+413
| if (hasChampion) { | ||
| const src = root.unwrap(championGenomeBuffer); | ||
| const encoder = root.device.createCommandEncoder(); | ||
| encoder.copyBufferToBuffer(src, 0, root.unwrap(ga.genomeBuffers[ga.current]), 0, src.size); | ||
| root.device.queue.submit([encoder.finish()]); | ||
| } |
Contributor
Author
There was a problem hiding this comment.
I'm not a fan of this but I think i prefer it to dispatching an entire compute shader to copy a couple of bytes
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 5 out of 7 changed files in this pull request and generated 2 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
apps/typegpu-docs/src/examples/algorithms/genetic-racing/track.ts
Outdated
Show resolved
Hide resolved
apps/typegpu-docs/src/examples/algorithms/genetic-racing/index.ts
Outdated
Show resolved
Hide resolved
Split Genome into InputLayer, DenseLayer and OutputLayer and replace flat weight vectors with mat4x4 matrices. Add FitnessArray and a fit compute shader plus a reduction to select the champion; wire the new fitness buffer into init/evolve flows. Update mutation/evolution helpers and simulation to evaluate the layered network and use the reworked sensor inputs.
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
reczkok
force-pushed
the
docs/genetic-racing
branch
from
9c5c3e4 to
c9fd29c
Compare
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 7 out of 9 changed files in this pull request and generated 1 comment.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
apps/typegpu-docs/src/examples/algorithms/genetic-racing/index.ts
Outdated
Show resolved
Hide resolved
iwoplaza
approved these changes
Mar 11, 2026
Collaborator
iwoplaza
left a comment
iwoplaza
left a comment
There was a problem hiding this comment.
Left a few nits, awesome example! And SO FAST 🚗🚗🚗🚗🚗🚗
| ); | ||
| }; | ||
|
|
||
| const evolveInputLayer = (a: d.Infer<typeof InputLayer>, b: d.Infer<typeof InputLayer>) => { |
Collaborator
There was a problem hiding this comment.
Suggested change
| const evolveInputLayer = (a: d.Infer<typeof InputLayer>, b: d.Infer<typeof InputLayer>) => { | |
| const evolveInputLayer = (a: d.InferGPU<typeof InputLayer>, b: d.InferGPU<typeof InputLayer>) => { |
| }); | ||
| }; | ||
|
|
||
| const evolveDenseLayer = (a: d.Infer<typeof DenseLayer>, b: d.Infer<typeof DenseLayer>) => { |
Collaborator
There was a problem hiding this comment.
Suggested change
| const evolveDenseLayer = (a: d.Infer<typeof DenseLayer>, b: d.Infer<typeof DenseLayer>) => { | |
| const evolveDenseLayer = (a: d.InferGPU<typeof DenseLayer>, b: d.InferGPU<typeof DenseLayer>) => { |
| return DenseLayer({ w: evolveMat4x4(a.w, b.w), bias: evolveVec(a.bias, b.bias) }); | ||
| }; | ||
|
|
||
| const evolveOutputLayer = (a: d.Infer<typeof OutputLayer>, b: d.Infer<typeof OutputLayer>) => { |
Collaborator
There was a problem hiding this comment.
Suggested change
| const evolveOutputLayer = (a: d.Infer<typeof OutputLayer>, b: d.Infer<typeof OutputLayer>) => { | |
| const evolveOutputLayer = (a: d.InferGPU<typeof OutputLayer>, b: d.InferGPU<typeof OutputLayer>) => { |
Comment on lines
+198
to
+207
| initLayout.$.genome[i] = Genome({ | ||
| h1: InputLayer({ | ||
| wA: randSignedMat4x4(), | ||
| wB: randSignedMat4x4(), | ||
| wC: randSignedMat4x4(), | ||
| bias: d.vec4f(), | ||
| }), | ||
| h2: DenseLayer({ w: randSignedMat4x4(), bias: d.vec4f() }), | ||
| out: OutputLayer({ steer: randSignedVec4(), throttle: randSignedVec4(), bias: d.vec2f() }), | ||
| }); |
Collaborator
There was a problem hiding this comment.
Is that really necessary? I thought we handled nested struct props
Suggested change
| initLayout.$.genome[i] = Genome({ | |
| h1: InputLayer({ | |
| wA: randSignedMat4x4(), | |
| wB: randSignedMat4x4(), | |
| wC: randSignedMat4x4(), | |
| bias: d.vec4f(), | |
| }), | |
| h2: DenseLayer({ w: randSignedMat4x4(), bias: d.vec4f() }), | |
| out: OutputLayer({ steer: randSignedVec4(), throttle: randSignedVec4(), bias: d.vec2f() }), | |
| }); | |
| initLayout.$.genome[i] = Genome({ | |
| h1: { | |
| wA: randSignedMat4x4(), | |
| wB: randSignedMat4x4(), | |
| wC: randSignedMat4x4(), | |
| bias: d.vec4f(), | |
| }, | |
| h2: { w: randSignedMat4x4(), bias: d.vec4f() }, | |
| out: { steer: randSignedVec4(), throttle: randSignedVec4(), bias: d.vec2f() }, | |
| }); |
| return hitT; | ||
| }; | ||
|
|
||
| const evalNetwork = (genome: d.Infer<typeof Genome>, a: d.v4f, b: d.v4f, c: d.v4f) => { |
Collaborator
There was a problem hiding this comment.
Suggested change
| const evalNetwork = (genome: d.Infer<typeof Genome>, a: d.v4f, b: d.v4f, c: d.v4f) => { | |
| const evalNetwork = (genome: d.InferGPU<typeof Genome>, a: d.v4f, b: d.v4f, c: d.v4f) => { |
| std.atomicMax(reductionLayout.$.packed, packed); | ||
| }); | ||
|
|
||
| const finalizeReductionPipeline = root.createGuardedComputePipeline((_x) => { |
Collaborator
There was a problem hiding this comment.
Why not a 0-dimension pipeline?
Suggested change
| const finalizeReductionPipeline = root.createGuardedComputePipeline((_x) => { | |
| const finalizeReductionPipeline = root.createGuardedComputePipeline(() => { |
Contributor
Author
There was a problem hiding this comment.
I recall having type errors when dispatching - will check
| const trackPipeline = root.createRenderPipeline({ | ||
| vertex: common.fullScreenTriangle, | ||
| fragment: trackFragment, | ||
| targets: { format: presentationFormat }, |
Collaborator
There was a problem hiding this comment.
Default
Suggested change
| targets: { format: presentationFormat }, |
| fragment: carFragment, | ||
| primitive: { topology: 'triangle-strip' }, | ||
| targets: { | ||
| format: presentationFormat, |
Collaborator
There was a problem hiding this comment.
Default
Suggested change
| format: presentationFormat, |
Comment on lines
+372
to
+377
| attribs: { | ||
| position: instanceLayout.attrib.position, | ||
| angle: instanceLayout.attrib.angle, | ||
| alive: instanceLayout.attrib.alive, | ||
| progress: instanceLayout.attrib.progress, | ||
| }, |
Collaborator
There was a problem hiding this comment.
Suggested change
| attribs: { | |
| position: instanceLayout.attrib.position, | |
| angle: instanceLayout.attrib.angle, | |
| alive: instanceLayout.attrib.alive, | |
| progress: instanceLayout.attrib.progress, | |
| }, | |
| attribs: instanceLayout.attrib, |
| const reductionEncoder = root.device.createCommandEncoder(); | ||
| reductionEncoder.clearBuffer(root.unwrap(reductionPackedBuffer)); | ||
| reductionPipeline.with(bg).with(reductionEncoder).dispatchThreads(population); | ||
| finalizeReductionPipeline.with(bg).with(reductionEncoder).dispatchThreads(1); |
Collaborator
There was a problem hiding this comment.
Suggested change
| finalizeReductionPipeline.with(bg).with(reductionEncoder).dispatchThreads(1); | |
| finalizeReductionPipeline.with(bg).with(reductionEncoder).dispatchThreads(); |
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.