Chrome 113 To Ship WebGPU By Default (phoronix.com) 43
While Chrome 112 just shipped this week and Chrome 113 only in beta, there is already a big reason to look forward to that next Chrome web browser release: Google is finally ready to ship WebGPU support. From a report: WebGPU provides the next-generation high performance 3D graphics API for the web. With next month's Chrome 113 stable release, the plan is to have WebGPU available out-of-the-box for this new web graphics API. Though in that version Google is limiting it to ChromeOS, macOS, and Windows... Yes, Google says other platforms like Linux will see their roll-out later in the year. The WebGPU API is more akin to Direct3D 12, Vulkan, and Metal compared with the existing WebGL being derived from OpenGL (ES). From Google's blog post: WebGPU is a new API for the web, which exposes modern hardware capabilities and allows rendering and computation operations on a GPU, similar to Direct3D 12, Metal, and Vulkan. Unlike the WebGL family of APIs, WebGPU offers access to more advanced GPU features and provides first-class support for general computations on the GPU. The API is designed with the web platform in mind, featuring an idiomatic JavaScript API, integration with promises, support for importing videos, and a polished developer experience with great error messages.
This initial release of WebGPU serves as a building block for future updates and enhancements. The API will offer more advanced graphics features, and developers are encouraged to send requests for additional features. The Chrome team also plans to provide deeper access to shader cores for even more machine learning optimizations and additional ergonomics in WGSL, the WebGPU Shading Language.
This initial release of WebGPU serves as a building block for future updates and enhancements. The API will offer more advanced graphics features, and developers are encouraged to send requests for additional features. The Chrome team also plans to provide deeper access to shader cores for even more machine learning optimizations and additional ergonomics in WGSL, the WebGPU Shading Language.
Comment removed (Score:3)
I hope it's off by default. (Score:4, Insightful)
"exposes modern hardware capabilities" = Hakt
Re: (Score:1)
Linux (Score:2)
Always a second class citizen, when is DEI going to work for us?
Re: (Score:2)
I think I'd be just as happy if this NEVER showed up on Linux. It sounds like a *really* bad idea.
Re: (Score:3)
Re:Linux (Score:5, Insightful)
There's a difference between wanting modern graphics, and being concerned that more open ended access to GPU could mean more than just modern graphics, including security risks for an area that has traditionally had local execution or higher level graphics library doing gatekeeping. One could imagine that it may turn out to be possible to use a shader to DMA in some memory from another piece of the browser process that was *ostensibly* sandboxed before. It wouldn't be the first time that in a rush to give web developers full access to the browser system resources that security problems were a result.
Re: (Score:2)
including security risks for an area that has traditionally had local execution or higher level graphics library doing gatekeeping.
You know your security concerns could be alleviated simply by looking at the pretty picture in TFA. Literally no one is talking about bypassing the existing local libraries doing gatekeeping here.
Re: (Score:2)
By traditional, I meant more restricted like WebGL, versus Vulkan where you have more arbitrary access.
The thing that is new here compared to current usual usage of Vulkan, is that a browser represents mixed trust domains in a single process. All those OS mechanisms in the pretty picture are designed to protect processes from each other, not so much a process from itself. A web browser in pursuit of efficiency mixes financial institutions and random crapware game sites all together in the same browser spa
Re: (Score:2)
Anything's possible, but the GPU enforces memory protection in its memory, and the IOMMU enforces memory protection between the GPU and the system, so it seems like the risk is fairly low and could be easily mitigated by having the GPU zero memory after release in its idle time.
Re: (Score:2)
the risk is fairly low and could be easily mitigated
Go tell that to Nintendo. The 3DS had gspwn. [3dbrew.org] Which used the fact that the GPU could bypass the kernel's memory protections. [3dbrew.org] This was actively used back during the early 3DS years to get homebrew (and "backups") working.
Even better, Nintendo failed to learn from their mistake with the 3DS and made it again on the early Switch models too. [switchbrew.org]
Not saying that PCs will have the exact same GPU issues mind you. Just giving a reminder that these things can happen in unexpected (and in some cases unpatchable) ways
Re: (Score:2)
The stress would be that in most GPU usage scenarios, the entire calling process is considered the same trust domain.
In a browser, they generally try to have different trust domains in the same process, which as far as the underlying OS and hardware are concerned are equally entitled to memory accesses. So those mechanisms are generally not expecting to have to protect memory of a process from that same process. So a 'fun game' tab knowing an exploit *might* somehow read memory unexpectedly, but it's the
Re: (Score:2)
I'm old enough to remember ActiveX, and how it allowed ordinary web pages to practically cripple the PCs on our college campus with crapware like Comet Cursor. You'd think someone would have thought running executable code right off the Internet was a terrible idea, but some corporate asshats went along with it anyway.
Never underestimate the damage that can be caused by reckless innovation. No, I don't want web pages to have low-level access to the GPU, let alone the ability to probe USB hardware or gauge
Re: (Score:2)
WebGL is a bit different than WebGPU in how open ended it is.
I don't know where you got that javascript or python code off the internet is more secure as a claim, that was never asserted.
Javascript persisted where Flash, Java, and ActiveX failed in the browser only because it modeled and implemented a security model where it was not allowed to interact with anything other than designated internet sites, some persistent storage specifically carved up for it, and some resources. E.g. it wasn't allowed to just
Re: Linux (Score:2)
All it needs is the approval button. Thereâ(TM)s things we need to do, like work on Jupyter notebooks with chatGPT. OpenAI cannot run our code, we need to run our own. Part of that equation is webAssembly, part of it is WebGPU. AI computing makes way more sense on the edge, especially for media.
Re: (Score:2)
It may make sense on the edge, but that does not necessarily mean it must be on a web browser. It's rather unfortunate that the only non-datacenter 'platform' that is given any consideration is a web browser.
Great news if you operate a sketchy crypto-mining (Score:4, Funny)
If it means we can go back to playing Quake Live at work, then fine. Otherwise, I can't really figure the draw.
Re: (Score:2)
Re: (Score:2)
This will not help put Quake in the browser. WebGL already exists. I'm not quite sure what point this serves other than crypto mining. I'm sure there are some things this can do that WebGL can't, but it's definitely at the edges of what's necessary.
Re: (Score:1)
Re: (Score:2)
Because we all prefer the web to be funded by advertising?
Dunno if you've seen the censorship-industrial complex that's spawned or not.
Re: Great news if you operate a sketchy crypto-min (Score:2)
You mean the trend which started fifteen years ago when independent blogs and forums decided to start censoring fuck and shit words so they wouldn't get kicked off of AdSense?
This is great. (Score:4, Funny)
Re: (Score:2)
Future advertisements are going to be AWESOME!
Just like Getting the News in HD -- as so many stations are fond of boasting, for some reason.
Re: (Score:2)
Nah, they're going to run ChatGPT training directly on your GPU... with the data they gobble from you.
Say goodbye to Windows 7. (Score:3)
Re: (Score:2)
Re: (Score:2)
Ok games. Browser games. But why? Most games in a browser suck anyway, so why?
Re: (Score:2)
I generally agree with being wary of the feature, however "Most games in a browser suck" as an argument is a bit of a self-fulfilling prophesy here. The browser games suck in part because they don't have access to do this stuff.
However, I do not like the continual 'as-a-servicing' of everything, since the biggest reason to make a non-sucky game browser based is to basically have supreme drm over not letting you run it locally.
Re: (Score:2)
Re: (Score:2)
When's the last time you played a game delivered through a browser? I haven't since the days of flash. It's going to be used for novelties, and for ads, and for ads for novelties.
Re: (Score:2)
Many homemade Game Boy games listed on Homebrew Hub [gbdev.io] run in the web-based emulator.
This Website Designed for Chrome! *flashing gif* (Score:1)
Re: (Score:2)
Just because you can... (Score:2)
I personally lack "courage" to ever use something like this. It seems unnecessarily reckless to a comical extent. The only thing I can think of that would be more reckless is implementing native sockets in browser.
Dodgy bastards (Score:1)
For Chromebooks' benefit? (Score:2)
I feel like this is probably something Google wanted in the browser because it has something to do with enabling more capabilities on a Chromebook?
(That's a scenario where users typically have very limited disk storage space to install local applications, and possibly lack user permissions to do so. But they might want to offer more graphics-intensive applications to them through a browser session.)
Otherwise, I agree with people that this just seems like it will cause more problems than it solves. Advertise
Re: (Score:2)
gamers will still prefer locally installed programs that load large maps and backgrounds from their SSD, vs trying to stream *everything* over the Internet connection.
The desires of gamers are irrelevant. Corporate demands piracy-proof*, game streaming-as-a-service TODAY!
*: OK, it's not entirely piracy-proof. But, the amount of work required to reconstruct a complete game from a recorded video stream is currently sufficient to deter most would be pirates.
Crypto scams (Score:2)
Finally, everyone can help the scammers mine some useless crypto!