LGTM to experiment unti M133 (inclusive), which would take this OT to 6 
milestones in total.

If you'd want to extend the experiment beyond that, I'd love to see 
improvements to the explainer, and/or meaning progress towards shipping 
(e.g. an early spec draft, test suite, WG discussions, etc). 

On Tuesday, October 15, 2024 at 6:51:37 PM UTC+2 Alan Baker wrote:

> Hi Alex,
>
> We received feedback based on the performance improvements the feature is 
> capable of delivering; however, those improvements might run afoul of 
> portability guard rails. I don't think testers would be specifically 
> waiting for these guard rails, but we think it would be good to get 
> feedback on them as it could affect the WG's decision surrounding what 
> should default to an error.
>
> The extension is mostly compatible, but it is possible that these new 
> guard rails cause existing shaders to fail to compile. We intentionally 
> left this out of the initial experiment to ensure we got a good signal on 
> the performance side of things. The new functionality does have controls to 
> revert to essentially the same behaviour as the current functionality with 
> some minor code changes.
>
> We felt it was a little premature to ship this feature since it is still 
> being designed at the WG. We're hoping that this extended functionality can 
> provide feedback for the WG to make the best decisions on the portability 
> aspects (mostly default error levels). We hope that by the time the this 
> extension runs out the WG will have progressed enough that we can send an 
> I2S for the full feature.
>
> Does that make sense? Admittedly, I'm not very familiar with this whole 
> process and appreciate any feedback you might have.
>
> Thanks,
> Alan
>
> On Fri, Oct 11, 2024 at 2:20 PM Alex Russell <slightly...@chromium.org> 
> wrote:
>
>> Hey Alan,
>>
>> Was there useful developer feedback from the first OT besides the lack of 
>> these features? Or are most potential testers waiting on the sidelines 
>> until these arrive?
>>
>> Also, will this be a compatible extension? That is, will code that worked 
>> in the previous version continue to work without changes? I'm not worried 
>> about a 7 month OT for something like this, but it's going to be tricky to 
>> go longer.
>>
>> The reason I ask is that If we *have *good signal from the already 
>> experimented subset and developers willing to try these extensions behind a 
>> flag and validate them, it might make good sense to just I2S now.
>>
>> Best,
>>
>> Alex
>>
>> On Wednesday, October 9, 2024 at 8:14:16 AM UTC-7 Alan Baker wrote:
>>
>>> Contact emailsalanba...@google.com
>>>
>>> Explainer
>>> https://github.com/gpuweb/gpuweb/blob/main/proposals/subgroups.md
>>>
>>> SpecificationNone
>>>
>>> Summary
>>>
>>> Adds subgroup functionality to WebGPU. Subgroup operations perform SIMT 
>>> operations to provide efficient communication and data sharing among groups 
>>> of invocations. These operations can be used to accelerate applications by 
>>> reducing memory overheads incurred by inter-invocation communication.
>>>
>>>
>>> Blink componentBlink>WebGPU 
>>> <https://bugs.chromium.org/p/chromium/issues/list?q=component:Blink%3EWebGPU>
>>>
>>> TAG reviewNone
>>>
>>> TAG review statusPending
>>>
>>> Chromium Trial NameWebGPUSubgroupsFeatures
>>>
>>> Origin Trial documentation link
>>> https://github.com/gpuweb/gpuweb/blob/main/proposals/subgroups.md
>>>
>>> WebFeature UseCounter namekWebGPUSubgroupsFeatures
>>>
>>> Risks
>>>
>>>
>>> Interoperability and Compatibility
>>>
>>> None
>>>
>>>
>>> *Gecko*: No signal
>>>
>>> *WebKit*: No signal
>>>
>>> *Web developers*: No signals
>>>
>>> *Other signals*:
>>>
>>> Security
>>>
>>> None. This feature exposes functionality for operations only within a 
>>> single draw/dispatch. Data is contained within a single users run.
>>>
>>>
>>> WebView application risks
>>>
>>> Does this intent deprecate or change behavior of existing APIs, such 
>>> that it has potentially high risk for Android WebView-based applications?
>>>
>>> None
>>>
>>>
>>> Goals for experimentation
>>>
>>>
>>>
>>> Reason this experiment is being extended
>>>
>>> The first portion of the origin trial focused on collecting performance 
>>> feedback data, but was an incomplete implementation of the proposal. In 
>>> particular, the implementation intentionally lacked the diagnostic controls 
>>> (and default errors) that serve as portability guard rails. We wish to 
>>> implement those controls (default values result in error in code that is 
>>> not provably portable) and collect feedback about how onerous the 
>>> portability restrictions are in order to provide feedback to the W3C WG. 
>>> The controls would allow non-portable code, but shader authors would have 
>>> to add code that opts into this behaviour.
>>>
>>>
>>> We wish to extend through M135.
>>>
>>>
>>> Ongoing technical constraints
>>>
>>> None
>>>
>>>
>>> Debuggability
>>>
>>> No special debugging is available for this feature as it resides solely 
>>> on the GPU.
>>>
>>>
>>> Will this feature be supported on all six Blink platforms (Windows, Mac, 
>>> Linux, ChromeOS, Android, and Android WebView)?No
>>>
>>> Is this feature fully tested by web-platform-tests 
>>> <https://chromium.googlesource.com/chromium/src/+/main/docs/testing/web_platform_tests.md>
>>> ?No
>>>
>>> Flag name on chrome://flagsNone
>>>
>>> Finch feature nameNone
>>>
>>> Non-finch justificationNone
>>>
>>> Requires code in //chrome?False
>>>
>>> Estimated milestones
>>> Origin trial desktop first 128
>>> Origin trial desktop last 131
>>> Origin trial extension 1 end milestone 135
>>> Origin trial Android first 128
>>> Origin trial Android last 131
>>> Origin trial WebView first 128
>>> Origin trial WebView last 131
>>>
>>> Link to entry on the Chrome Platform Status
>>> https://chromestatus.com/feature/5126409856221184?gate=5073697141161984
>>>
>>> Links to previous Intent discussionsIntent to Experiment: 
>>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CABg%2BrB67gB5gUeDVPaMZ7fZwmeARTUsizs1BTXFXKGpJ%3DpF1TQ%40mail.gmail.com
>>>
>>>
>>> This intent message was generated by Chrome Platform Status 
>>> <https://chromestatus.com/>.
>>>
>>

-- 
You received this message because you are subscribed to the Google Groups 
"blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to blink-dev+unsubscr...@chromium.org.
To view this discussion on the web visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/a904792d-f652-46f6-b8bf-2f2d0ed4ca9fn%40chromium.org.

Reply via email to