Reads pretty much like an advertisement.

Nothing that interesting; 2D game engine Construct moved to WebGPU, and reaps some obvious performance benefits. Great.

Obvious only if you already know what WebGPU is and how it compares to WebGL.

It's an announcement for a major update with relevant benchmarks and some basic technical details. What more do you want?

You might find this older post more interesting, more technical. Just keep in mind this blog post is 3 years old:

https://www.construct.net/en/blogs/ashleys-blog-2/webgl-webg...

2D blitting is just not that exciting, no matter how it's done.

It seemed exciting to me! Maybe gatekeeping technical knowledge is not something we want to do...

The large, distracting menus are quite exciting.

I think 2d blitting is plenty exciting and this article looks like a great fit for content on hackernews.

At least for me, it's an awareness thing.

"oh cool webgpu is slowly starting to appear in production environments. Positive steps towards the technology getting broad adoption".

I don't really even take note of the company in this instance.

No video?

They provided links to the benchmarks so you can try running them yourself. I feel like a prerecorded video would add little value to this announcement.

https://downloads.scirra.com/labs/perf/opacitythrash-webgl/

https://downloads.scirra.com/labs/perf/opacitythrash-webgpu/

https://downloads.scirra.com/labs/perf/colorthrash-webgl/

https://downloads.scirra.com/labs/perf/colorthrash-webgpu/

I've happy about the recent stabilization of WebGPU in Chrome, and excited for the rest of the browsers to follow suit.

I wanted to start experimenting with it, and attempted to do so with Deno (which is what I've been developing in the past few projects), and I'm sad to find out that they recently removed WebGPU from its interface due to slower start up times and larger binary sizes [0][1] :/ I hope they re-enable it soon so that I don't have to mess around with tsc and webpack again...

[0] https://stackoverflow.com/questions/76561043/deno-bundle-pro... [1] https://github.com/denoland/deno/pull/18094

I wouldn't say Chromium's WebGPU is stable just yet. It still doesn't run on my Linux system anyway. Mind you that both the Dawn and WGPU c libraries work on this system.

It is kind of ironic that with so many Linux based stuff and two operating systems, the Mountain View gnomes have decided to support Windows and Mac first.

Kind of tells where they see business value to deploy first.

I didn't know what Construct was so I went to the homepage for clues. That intro spinning cube demo runs at a solid 4 fps with graphical glitches on a not exactly slow phone (Pixel 7), give or take, and is a quite terrible introduction for a "Cutting-edge technology that runs right inside your browser like magic."

Then again the entire section on "game engine performance matters" only compares the performance of "JavaScript" (V8? Safari? Who knows!) vs. Game Maker Language and doesn't at any point compare anything about the performance of the actual game engine.

But it is cool that they're switching to WebGPU. Should make native app ports more viable to drastically improve performance and resource utilization

FWIW that intro cube is not implemented with their engine. It's just 4 video embeds animated with plain CSS 3D transforms, so performance should depend entirely on browser optimization on your device.

Seems Firefox hates it for some reason

Weird, Firefox on my 7a seems to handle it qute fine.

The intro runs fine on my nondescript phone of a few years ago. I would guess it's a browser issue.

It lags on my laptop with Firefox, is a black cube with Chrome, and has a lot of hiccups on my phone (one+ 8 pro)

After they switched to a monthly/annual subscription fee with the release of construct 3, I pretty much threw in the towel and switched over to Gdevelop.

https://github.com/4ian/GDevelop

Open source, completely free, and I can run it as a native application on my computer versus a weird web app. The idea that my game is basically tied to a SaaS is just not OK for me.

I'm not an expert in this stuff, but it seems a bit premature to go to production with anything WebGPU at this point. Chrome made it available by default just a couple months ago. It's one thing if it's like a new CSS property or something, but WebGPU is pretty complex system that I think will take a long time to mature IMO.

The API isn't going to change. This company will adopt it eventually, it's a good idea to take it for a spin.