feat: CPU implementation of std.transpose#2737
Open
iwoplaza wants to merge 1 commit into
Open
Conversation
iwoplaza
force-pushed
the
feat/std-transpose-on-cpu
branch
from
July 16, 2026 21:12
c52a42e to
53d569a
Compare
iwoplaza
marked this pull request as ready for review
July 16, 2026 21:13
|
pkg.pr.new packages benchmark commit |
Bundle size comparison (
|
| 🟢 Decreased | ➖ Unchanged | 🔴 Increased (max 0.02%) | ❔ Unknown |
|---|---|---|---|
| 0 | 239 | 83 | 0 |
import { ... } in PR vs import * as ... in PR (is the library tree-Shakeable?):
| Test | tsdown |
|---|---|
| tgpu_init.ts | 258.91 kB ( |
| tgpu_initFromDevice.ts | 258.40 kB ( |
| tgpu_resolve.ts | 164.60 kB ( |
| tgpu_resolveWithContext.ts | 164.54 kB ( |
| tgpu_bindGroupLayout.ts | 68.97 kB ( |
| tgpu_mutableAccessor.ts | 65.97 kB ( |
| tgpu_accessor.ts | 65.96 kB ( |
| tgpu_privateVar.ts | 65.31 kB ( |
| tgpu_workgroupVar.ts | 65.30 kB ( |
| tgpu_const.ts | 64.55 kB ( |
| tgpu_fn.ts | 38.14 kB ( |
| tgpu_fragmentFn.ts | 38.14 kB ( |
| tgpu_vertexFn.ts | 37.96 kB ( |
| tgpu_computeFn.ts | 37.66 kB ( |
| tgpu_vertexLayout.ts | 26.79 kB ( |
| tgpu_comptime.ts | 14.48 kB ( |
| tgpu_unroll.ts | 1.66 kB ( |
| tgpu_slot.ts | 1.54 kB ( |
| tgpu_lazy.ts | 1.19 kB ( |
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 [0.88, 1.77, 4.00, 6.06, 6.72, 11.78, 21.16, 23.25]
line [0.93, 1.87, 4.32, 6.09, 7.03, 12.35, 21.96, 24.42]
line [0.91, 1.85, 3.80, 6.27, 7.28, 11.96, 21.44, 22.58]
---
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 [0.30, 0.46, 0.70, 0.83, 1.13, 1.18, 1.43, 1.48]
line [0.28, 0.51, 0.68, 0.82, 1.11, 1.21, 1.42, 1.52]
line [0.29, 0.51, 0.69, 0.83, 1.17, 1.26, 1.42, 1.55]
---
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 [0.73, 2.15, 4.16, 6.67, 11.81, 24.15, 51.52, 105.95]
line [0.83, 1.92, 3.35, 6.38, 12.42, 25.83, 54.02, 109.76]
line [0.79, 1.86, 3.65, 5.95, 12.19, 24.69, 52.84, 108.41]
|
iwoplaza
force-pushed
the
feat/std-transpose-on-cpu
branch
from
July 16, 2026 21:15
53d569a to
fb18dab
Compare
There was a problem hiding this comment.
Pull request overview
Adds a CPU-side implementation for std.transpose so matrix transposition works in JS execution, while still generating a WGSL transpose(...) call when used in GPU-transpiled code.
Changes:
- Implemented
std.transposeCPU behavior format2x2f,mat3x3f, andmat4x4fviaWORKAROUND_getSchema(...). - Added unit tests covering JS transposition results for 2x2/3x3/4x4 matrices.
- Added a codegen test asserting WGSL output uses the builtin
transpose(...)call.
Reviewed changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| packages/typegpu/src/std/numeric.ts | Implements CPU transpose for square float matrices and wires it into dualImpl while keeping WGSL codegen. |
| packages/typegpu/tests/std/transpose.test.ts | Adds coverage for CPU results and validates GPU codegen emits transpose(foo) in WGSL. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
![pullfrog[bot]](https://avatars.githubusercontent.com/in/1768019?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes — implemented the missing host-side CPU/JS std.transpose and added tests covering all currently supported square matrix sizes.
- Added
cpuTransposeinpackages/typegpu/src/std/numeric.tsto replace the placeholder error message. - Wired
unaryIdentitySignaturefor runtime/GPU type resolution, consistent with other unary builtins likeabs. - Added unit tests for
mat2x2f,mat3x3f, andmat4x4fhost transpositions plus atgpu.resolvesnapshot for the GPUtranspose()builtin path.
Kimi K2 (free via Pullfrog for OSS) ｜ 𝕏
Contributor
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes — incremental — added explicit toString(): string declarations to the matrix instance interfaces and removed the now-unnecessary typescript/no-base-to-string lint override in niceStringify. The original cpuTranspose implementation and tests remain unchanged.
packages/typegpu/src/data/wgslTypes.tsnow declarestoString(): stringonmat2x2,mat3x3, andmat4x4interfaces, matching the runtime method already provided by the concrete matrix classes.packages/typegpu/src/shared/stringify.tsdrops the oxlint disable comment because matrices are no longer treated as implicitly usingObject.prototype.toString.- Re-verified:
npx vitest run tests/std— 374 passed; oxlint on changed files — clean.
Kimi K2 (free via Pullfrog for OSS) ｜ 𝕏
cieplypolar
reviewed
Jul 17, 2026
| }); | ||
|
|
||
| export const transpose = dualImpl<<T extends AnyMatInstance>(e: T) => T>({ | ||
| function cpuTranspose<T extends AnyMatInstance>(value: T): T { |
Collaborator
There was a problem hiding this comment.
I would hardcode it for every size in {2, 3, 4}, for example 2x2 transposition
const schema = WORKAROUND_getSchema(value);
const transposed = schema() as T;
const src = value.columns;
const dst = transposed.columns;
dst[0]![0] = src[0]![0]!;
dst[0]![1] = src[1]![0]!;
dst[1]![0] = src[0]![1]!;
dst[1]![1] = src[1]![1]!;
return transposed;
Collaborator
Author
There was a problem hiding this comment.
For the sake of performance?
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.
In addition, the TS type system is now made aware that matrices have a custom
toStringmethod, so ignoring the lint rule is no longer necessary.