Remove MLDeviceType #809

zolkis · 2025-01-27T16:33:00Z

Resolves #749.
Addresses #302, #350.
See also the device selection explainer.

This is a minimal change for now. It may require changing prose or adding more details / instructions in related sections/algorithms.

Preview | Diff

inexorabletash · 2025-01-27T17:48:30Z

@zolkis can you use the keyword "Resolves #749" in the PR description? This creates a visible link between the issue and PR.

Also... would you consider #302 addressed by this PR?

inexorabletash

Overall LGTM, lots of nitpicks.

index.bs

anssiko · 2025-01-30T16:39:42Z

@zolkis I believe this PR also addresses #350 by making MLContext device agnostic.

index.bs

anssiko · 2025-02-04T18:24:21Z

@fdwr commented:

Thoughts? Do we delete MLDeviceType for now and then resurrect it after deciding how to query which kinds of devices a context can support? Do keep it but remove the context creation parameter?

Since enums are not web-exposed, we could leave the MLDeviceType enum in place (for later reuse/repurpose) but remove it from the MLContextOptions dictionary. In addition could update this note to add that a mechanism to query devices a context can support is being considered.

I created a new issue for discussing such a query mechanism: #815

or it might be okay to use CPU, but the app could load a different model that's more CPU-friendly, if it knew that was the case.

@zolkis I think this should be added to the device selection explainer as a new use case.

index.bs

@fdwr

* Ensure object creation specifies the realm "Realm" is an ECMAScript concept best explained in https://html.spec.whatwg.org/multipage/webappapis.html#realms-and-their-counterparts Newly created JS objects must be associated with a Realm; while older specs didn't do this explicitly, best practice is to be explicit about this, especially for steps running "in parallel", or in algorithms separate from method steps. Do so! This also adds lint tests to try and catch future violations. Note that dictionaries (e.g. MLOperatorDescriptor) are Infra "ordered maps" it the body of spec algorithms, not JS objects, so they don't have a realm. Conversion to a JS object when returning a dictionary to script is handled by WebIDL bindings logic. Also note that DOMExceptions, either thrown or as promise rejection values, are not given a realm. This is a known issue across all web specs and is tracked in whatwg/webidl#135. Resolves webmachinelearning#793. * Don't double-init realm; and don't need realm for dicts * Variable name improvement from @fdwr

Signed-off-by: Zoltan Kis <[email protected]> Improve text on device selection Co-authored-by: Joshua Bell <[email protected]> Improve text on fingerprinting Co-authored-by: Joshua Bell <[email protected]> Improve text on underlying device selection Co-authored-by: Joshua Bell <[email protected]> Address review comments from Ningxin Fix RFC2119 issue for Note Signed-off-by: Zoltan Kis <[email protected]>

zolkis · 2025-02-06T15:13:57Z

Apologies for the forced pushes. Squashed these simple changes into one commit, but was interleaved with others from the main, so it needed rebase, then reorder/merge, then squash/merge. Now it should be clean to merge.

index.bs

anssiko · 2025-02-06T16:27:32Z

@zolkis thanks for addressing the review feedback. I added a few suggestions for the open points and marked as resolved those conversations that were addressed. Please double-check.

To allow for adequate review time, let's keep this PR open until our next call, 2025-02-13.

Co-authored-by: Anssi Kostiainen <[email protected]>

huningxin

LGTM!

anssiko · 2025-02-13T16:36:24Z

Per our discussion on WebML WG Teleconference – 13 February 2025, this PR is ready to merge. Thank you everyone!

SHA: 9c00304 Reason: push, by anssiko Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

This commit comprehensively updates the device selection explainer to reflect the latest discussions, API changes, and community feedback. Key changes include: - Updated Introduction and History sections to accurately reflect the removal of `MLDeviceType` from `MLContextOptions` (following PR webmachinelearning#809) and the shift towards hint-based, implementation-led device selection. - Clarified the impact of key issues (webmachinelearning#749, webmachinelearning#302, webmachinelearning#350) and PRs (webmachinelearning#809, webmachinelearning#824, webmachinelearning#855) on the device selection strategy. - Ensured the "Key use cases and requirements" section aligns with the current API, incorporating the device preference use cases from PR webmachinelearning#855. - Updated JavaScript examples in "Scenarios, examples, design discussion" to be consistent with the current API, marking future/hypothetical features (like `opSupportLimitsPerDevice()` and a `fallback` option) with explanatory notes. - Added new open questions based on recent discussions (e.g., issue webmachinelearning#836, PR webmachinelearning#854 regarding querying actual device usage). - Refined the "Background thoughts" section, particularly the "Example Hardware Selection Guide," adding an editor's note about ongoing discussions (PR webmachinelearning#860). - Corrected the "Considered alternatives" and "Minimum Viable Solution" sections to accurately represent the current and past approaches. - Updated the "Next Phase Device Selection Solution" to clarify the status of proposals like `querySupport` (issue webmachinelearning#815) and the investigation of `graph.devices` (issue webmachinelearning#836, PR webmachinelearning#854). - Performed a full proofread, correcting grammar, typos, and markdown formatting for improved clarity and consistency throughout the document.

This commit comprehensively updates the device selection explainer to reflect the latest discussions, API changes, and community feedback. Key changes include: - Updated Introduction and History sections to accurately reflect the removal of `MLDeviceType` from `MLContextOptions` (following PR #809) and the shift towards hint-based, implementation-led device selection. - Clarified the impact of key issues (#749, #302, #350) and PRs (#809, #824, #855) on the device selection strategy. - Ensured the "Key use cases and requirements" section aligns with the current API, incorporating the device preference use cases from PR #855. - Updated JavaScript examples in "Scenarios, examples, design discussion" to be consistent with the current API, marking future/hypothetical features (like `opSupportLimitsPerDevice()` and a `fallback` option) with explanatory notes. - Added new open questions based on recent discussions (e.g., issue #836, PR #854 regarding querying actual device usage). - Refined the "Background thoughts" section, particularly the "Example Hardware Selection Guide," adding an editor's note about ongoing discussions (PR #860). - Corrected the "Considered alternatives" and "Minimum Viable Solution" sections to accurately represent the current and past approaches. - Updated the "Next Phase Device Selection Solution" to clarify the status of proposals like `querySupport` (issue #815) and the investigation of `graph.devices` (issue #836, PR #854). - Performed a full proofread, correcting grammar, typos, and markdown formatting for improved clarity and consistency throughout the document. Signed-off-by: Zoltan Kis <[email protected]> With thanks to Jules for proofreading: Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>

inexorabletash added the device selection label Jan 27, 2025

inexorabletash reviewed Jan 27, 2025

View reviewed changes

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

index.bs Show resolved Hide resolved

dontcallmedom mentioned this pull request Jan 30, 2025

[wg/webmachinelearning] Web Machine Learning Working Group Charter w3c/strategy#488

Closed

1 task

plehegar mentioned this pull request Jan 30, 2025

Need to understand how WebNN supports implementation that involves multiple devices and timelines #350

Closed

fdwr reviewed Jan 31, 2025

View reviewed changes

index.bs Show resolved Hide resolved

anssiko mentioned this pull request Feb 4, 2025

Query supported devices before graph compilation #815

Open

mwyrzykowski approved these changes Feb 4, 2025

View reviewed changes

mwyrzykowski reviewed Feb 4, 2025

View reviewed changes

index.bs Outdated Show resolved Hide resolved

fdwr approved these changes Feb 5, 2025

View reviewed changes

index.bs Outdated Show resolved Hide resolved

huningxin reviewed Feb 6, 2025

View reviewed changes

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

zolkis force-pushed the device-selection-refactoring branch 4 times, most recently from 0550c98 to b10fef0 Compare February 6, 2025 15:07

inexorabletash and others added 2 commits February 6, 2025 17:10

zolkis force-pushed the device-selection-refactoring branch from b10fef0 to 7ce5a42 Compare February 6, 2025 15:11

anssiko approved these changes Feb 6, 2025

View reviewed changes

index.bs Outdated Show resolved Hide resolved

index.bs Outdated Show resolved Hide resolved

zolkis and others added 2 commits February 6, 2025 18:51

Update note about context options

e5e603f

Co-authored-by: Anssi Kostiainen <[email protected]>

Improve bailout step for context creation.

fee790b

Co-authored-by: Anssi Kostiainen <[email protected]>

huningxin approved these changes Feb 8, 2025

View reviewed changes

mwyrzykowski mentioned this pull request Feb 13, 2025

MLContextOptions.deviceType seems unnecessary outside of conformance testing #749

Closed

anssiko merged commit 9c00304 into webmachinelearning:main Feb 13, 2025
2 checks passed

github-actions bot added a commit that referenced this pull request Feb 13, 2025

Device selection - Remove MLDeviceType (#809)

842ca5a

SHA: 9c00304 Reason: push, by anssiko Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

anssiko mentioned this pull request Mar 11, 2025

Add NPU support and updating URL arguments for developer preview demos webmachinelearning/webnn-samples-test-framework#3

Merged

zolkis deleted the device-selection-refactoring branch May 27, 2025 12:02

Uh oh!

Remove MLDeviceType #809

Remove MLDeviceType #809

Uh oh!

Conversation

zolkis commented Jan 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

inexorabletash commented Jan 27, 2025

Uh oh!

inexorabletash left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

anssiko commented Jan 30, 2025

Uh oh!

Uh oh!

anssiko commented Feb 4, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

zolkis commented Feb 6, 2025

Uh oh!

Uh oh!

Uh oh!

anssiko commented Feb 6, 2025

Uh oh!

huningxin left a comment

Choose a reason for hiding this comment

Uh oh!

anssiko commented Feb 13, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

zolkis commented Jan 27, 2025 •

edited

Loading