Bump

Here's a video I found that explains the purpose of WebGPU for those confused: https://www.youtube.com/watch?v=oIur9NATg-I

Sounds like an interesting idea. Does anybody know about the actual process of swapping this all around? It seems complicated. If not, then it's definitely something worth investing in.

-Valtren- wrote:

Support, I often work with GPU intensive projects. And if this was implemented, it would likely be cause for 4.0.

(Replying to all of the messages talking about versions of scratch, not just this one)
I doubt they'll ever brand an update to scratch as an X.0 update. 1.0 was initial release, 2.0 revamped everything, and 3.0 once again revamped everything. It would be hard to make a redesign to scratch again, as it's already quite modern. Just like the dozens of updates since 2019, I'm sure they'll probably just give it a name, and keep it like that.

This is a really cool idea, support. Helps to make everything faster and I'd like to see it in a potential 4.0 release

There could also be a toggle-able option to use WebGPU or WebGL 1 in the case people want to use Scratch but have low power CPUs.

XCartooonX wrote:

(#63)
This is a really cool idea, support. Helps to make everything faster and I'd like to see it in a potential 4.0 release

There could also be a toggle-able option to use WebGPU or WebGL 1 in the case people want to use Scratch but have low power CPUs.

Low power GPU's, not CPUs. The GPU is what WebGPU accesses, not the CPU, and the CPU has minimal effect on WebGPU performance.

anyways there should be a message if webgpu is disabled
(and according to the browser you use it should tell you where to enable)
(if the browser doesn’t support webgpu then just say that its disabled)

p-p-p-p-p-p-p-p-p-p- wrote:

(#62)
Sounds like an interesting idea. Does anybody know about the actual process of swapping this all around? It seems complicated. If not, then it's definitely something worth investing in.

-Valtren- wrote:

Support, I often work with GPU intensive projects. And if this was implemented, it would likely be cause for 4.0.

(Replying to all of the messages talking about versions of scratch, not just this one)
I doubt they'll ever brand an update to scratch as an X.0 update. 1.0 was initial release, 2.0 revamped everything, and 3.0 once again revamped everything. It would be hard to make a redesign to scratch again, as it's already quite modern. Just like the dozens of updates since 2019, I'm sure they'll probably just give it a name, and keep it like that.

It would be a good thing to call it Scratch 4, as it would redo a major part of the engine, even if it's under the hood. Scratch 2 was the Flash rewrite, Scratch 3 was the JS+WebGL rewrite, and Scratch 4 will be the WebGPU rewrite.

mumu245 wrote:

p-p-p-p-p-p-p-p-p-p- wrote:

(#62)
Sounds like an interesting idea. Does anybody know about the actual process of swapping this all around? It seems complicated. If not, then it's definitely something worth investing in.

-Valtren- wrote:

Support, I often work with GPU intensive projects. And if this was implemented, it would likely be cause for 4.0.

(Replying to all of the messages talking about versions of scratch, not just this one)
I doubt they'll ever brand an update to scratch as an X.0 update. 1.0 was initial release, 2.0 revamped everything, and 3.0 once again revamped everything. It would be hard to make a redesign to scratch again, as it's already quite modern. Just like the dozens of updates since 2019, I'm sure they'll probably just give it a name, and keep it like that.

It would be a good thing to call it Scratch 4, as it would redo a major part of the engine, even if it's under the hood. Scratch 2 was the Flash rewrite, Scratch 3 was the JS+WebGL rewrite, and Scratch 4 will be the WebGPU rewrite.

However that's just an under the hood update, 2.0 and 3.0 were both site wide updates.

p-p-p-p-p-p-p-p-p-p- wrote:

(#67)

mumu245 wrote:

p-p-p-p-p-p-p-p-p-p- wrote:

(#62)
Sounds like an interesting idea. Does anybody know about the actual process of swapping this all around? It seems complicated. If not, then it's definitely something worth investing in.

-Valtren- wrote:

Support, I often work with GPU intensive projects. And if this was implemented, it would likely be cause for 4.0.

(Replying to all of the messages talking about versions of scratch, not just this one)
I doubt they'll ever brand an update to scratch as an X.0 update. 1.0 was initial release, 2.0 revamped everything, and 3.0 once again revamped everything. It would be hard to make a redesign to scratch again, as it's already quite modern. Just like the dozens of updates since 2019, I'm sure they'll probably just give it a name, and keep it like that.

It would be a good thing to call it Scratch 4, as it would redo a major part of the engine, even if it's under the hood. Scratch 2 was the Flash rewrite, Scratch 3 was the JS+WebGL rewrite, and Scratch 4 will be the WebGPU rewrite.

However that's just an under the hood update, 2.0 and 3.0 were both site wide updates.

The reason they got a new version number is that they rewrote a lot of things. 1.1 was also a site-wide update.

Bump

Bringing this topic back up again

The speed limit (on computers with good graphics hardware) usually comes from the Scratch interpreter and not from WebGL, that's why TurboWarp can speed up projects

nembence wrote:

(#71)
The speed limit (on computers with good graphics hardware) usually comes from the Scratch interpreter and not from WebGL, that's why TurboWarp can speed up projects

But even if it could just speed up drawing lines with the pen it would be an awesome improvement.

Bringing this one back up again.

this was originally closed because the op @ajstakeboarder
didn’t agree with the suggestion because webgpu wouldn’t bring any “performance benefits”

source:
https://scratch.mit.edu/discuss/post/7924695/

idk why it’s up again

also I was kinda dumb for mentioning installing vulkan as it was installing the runtime and the sdk
and NOT the actual support