Chrome will use gpu to render pages | Hacker News Reader

Chrome will use gpu to render pages

blog.chromium.org

91 pointscatalinist47 comments

Open articleView on HN

Not trying to detract from Chrome here, but IE9 should have a similar ability:

http://blogs.msdn.com/ie/archive/2010/04/09/Benefits-of-GPU-...

Not trying to detract from IE9 here, but Opera already has a similar ability:

http://my.opera.com/core/blog/2009/02/04/vega

The interesting parts of the article are the implementation details about the sandbox.

Opera's hardware acceleration is vaporware. It has not been included in any release of Opera, despite that article being over a year old now.

It's not vaporware. It's been in Opera since 10.5. It isn't being used at all but it's still in the release.

The thing about Opera is, it only has around 2.5% of the market. IE9 will probably have over 30% of the browser market within two years, which is a major factor when thinking about any web development, especially with HTML5 features like canvas.

The reason it took longer for chrome is that they are trying to sandbox everything, including what goes through the GPU which according to the chrome devs is very difficult and painful. They could have gotten the whole thing to render on the GPU a long time ago if it wasn't for the sandboxing.

Not to mention doing it on all 3 platforms as well as translating OpenGL to Direct3D on the fly using ANGLE.

"took longer"? They are likely to beat IE9 to official release.

Has. A preview is available for download now: http://ie.microsoft.com/testdrive/

Good - Opera, Firefox, and IE9 are all aready going this route. For Chromium not to support this would be a big thorn in their side, especially with their plans to support native code for extensions (aka, "Chrome Apps").

Looking forward to this - I frequently scale web pages (my vision isn't great) and it bugs me this common activity still isn't smooth in 2010.

Have you tried an iPad? For some reason Mobile Safari is still far better at scaling pages than desktop Safari.

As exciting as it is for IE9 to enter the next-gen browser wars in earnest I wonder if they are already too late.

IE releases are still tied to the years long windows ship cycle, which is a significant disadvantage when all your competitors are on months long cycles. Saddled with such a huge OODA disadvantage as well as all the strategy-tax BS from MS is there any hope?

By the time IE9 is actually out Chrome may well have hardware accelerated everything, an LLVM client/plugin model, javascript that runs faster than Java, and client-side Erlang support (or who knows what), with Firefox not far behind. It'll be interesting to see how this plays out.

"IE releases are still tied to the years long windows ship cycle, ..."

I'm pretty sure I get notifications for Windows Updates when there's a new version of IE. And even if not, I've somehow gotten IE 7 or 8 on an XP install, so it's no different than when Mozilla or Google announce a new release. If you want it, you get it.

If you want it, you get it.

The people who want it are irrelevant. So are the people who don't want it. Both are minorities compared to the people who don't care, and just pick the default option. Chrome's default is to silently upgrade, and light a tiny orange dot when you need to restart.

"Chrome's default is to silently upgrade, and light a tiny orange dot when you need to restart."

I ended up with IE8 because I ran Windows Update, not because I had some desire to get a new IE.

My guess is that most Windows users are going to let the updater run, and get new versions of IE.

I ended up with IE8 because I ran Windows Update, not because I had some desire to get a new IE.

Exactly. Chrome updates faster because it does it with zero user input. It just happens whenever spare resources are available and politely lets you know when it's done. IE8 needs you to run Windows Update, which a significant number of users don't want the hassle.

" IE8 needs you to run Windows Update, which a significant number of users don't want the hassle."

I think we're each arguing from incompatible anecdotes.

But when it comes to browsers, IE is the default option, so that shouldn't hurt them, really.

IE9 is a lot friendlier to a system admin trying to install on hundreds of computers. The default Chrome download doesn't even install system wide. It would help if Google has a simple MSI package that did a systemwide install and allowed for updates from a local server, not calling Google all the time.

The automatic and frequent update system is one thing which makes Chrome very good. It can pretty much be assumed that all users connected to the internet would be atleast on the latest stable version.

Actually, in a place with a good internal network and a lower speed external connection, it makes Chrome very bad. Taking up bandwidth downloading the same thing for multiple computers is bad. Windows and OS X allow one download to be propagated to multiple machines. This is an efficient model that allows reporting on which machines didn't get the update (thus making it a lot more likely all machines will be updated).

We all don't have the same internet connections that Mountain View has.

> We all don't have the same internet connections that Mountain View has.

Luckily, Chrome use differential updates. This means each update is a tiny binary patch. If you combine them all on your network, it's still a smaller update then one single IE update of 250MB deployed to each machines.

1 250mb internet download vs. 4-8mb for a couple hundred machines on the internet line is a big difference. Particularly since I can plan when the big download happens (and update) versus all the small downloads that happen in a lab at the same time because of a class starting. Students with the first class of the day don't feel terribly happy about being the designated updater of some company's software.

> 1 250mb internet download vs. 4-8mb

Make that a few kb. And last time I checked at my work, admin updates usually happen during work hours and affects people working in the "lab". Also, it's 1 250mb that is sometimes installed on each machine on the network, so way worse than a few kb.

Well, your admin is doing it wrong if updates are happening during the day. So, no people bothered, reports, and less internet usage is better.

Your admin is doing it wrong if updates of a few kb are clogging his network. Also, small labs can't afford to pay admins at night, and big labs surely can afford a few kb transfers on their networks. We're talking about gigabytes networks and admin can always use cache proxy so that chrome only taps the net once for the tiny kb update.

I highly suspect that you are projecting the experience of other auto-updating software (such as Adobe Reader) onto Chrome. Chrome's auto-updates are incredibly efficient and very non-intrusive.

The problem with allowing admins to manage updates is that any new version which breaks an internal app in the slightest way would be a reason to avoid updating.

This is what has given IE6 its long life in the corporate world. While this is a pragmatic decision, it can hurt badly in the long run as we all know.

So, it's preferable to have a company not able to use its apps to get business done then have a non-updated browser?

IE releases are still tied to the years long windows ship cycle

Are you saying that IE 9 won't ship until Windows 8 ships? Because what I'v read shows IE 9 shipping in 2011 and Windows 8 in 2012.

If we're going to jump into that line of thought: Vista shipped 3 years late and with half of the "cool" features stripped out. I think the IE 9 team has something to prove now, and it's obvious Microsoft is making a push in making IE not suck.

It does seem too late for them to pull off something like they did with IE6 versus Netscape, but I could see their genuine efforts with IE9 keeping them from getting thrown to the wolves by average users. Many people still using IE regularly probably don't care (or even know) what browser they are using, so if the websites they visit work, they'll probably never think of changing and just take whatever MS puts out. I think IE9 could come in time to at least maintain their total browser share of around 50% over the next five or so years.

Most people have no idea what a browser is. As long as MS ship IE as the default browser without choice, they will have a massive competitive advantage and keep the majority of users.

In European versions of Windows, you now get a choice between which web browser to use. That's about as fair as it's going to get.

IE still has 60% market share, and it's holding steady. IE9 is going to be one of the most important products Microsoft ever ships. It will be compatible enough with web standards, and offer enough performance, that it should allow developers to build more sophisticated apps based on open standards. Html5, css3 and fast Javascript will soon be a given.

Holding steady? How do you figure that, exactly?

http://en.wikipedia.org/wiki/Usage_share_of_web_browsers#His...

I figured that because between June and July, IE marketcshare increased by .4%

http://marketshare.hitslink.com/browser-market-share.aspx?qp...

> IE still has 60% market share, and it's holding steady.

You mean: "has been dropping steadily".

> IE9 is going to be one of the most important products Microsoft ever ships.

I've heard that about '95, '98, 2000, XP, Vista and windows 7 as well as a whole bunch of products. They can't all be 'one of the most important products Microsoft ever ships'.

> It will be compatible enough with web standards, and offer enough performance, that it should allow developers to build more sophisticated apps based on open standards. Html5, css3 and fast Javascript will soon be a given.

They already are, just not in the IE world. That's the reason for line one above, Microsoft has been playing catch-up.

IE had 95%. Things change, sometimes things change quickly. Everyone who is capable of using facebook is capable of installing chrome or firefox. If IE truly fell behind, if a new killer web-app/site comes along that didn't work in IE the marketshare numbers would change overnight.

> if a new killer web-app/site comes along that didn't work in IE the marketshare numbers would change overnight.

That could be the chrome web store with all the cool webgl 3d games (IE9 won't support webgl and google is sure going to take advantage of this).

Matt Cutts tweeted a design document about the technical details

https://sites.google.com/a/chromium.org/dev/developers/desig...

Also Flash 11 is about to support 3D and it looks serious

http://news.cnet.com/8301-30685_3-20009940-264.html

Unity3D is at a turning point.

Yeah, it is at a turning point, it's about to become plugin-less using Native Client.

http://blogs.unity3d.com/2010/05/19/google-android-and-the-f...

It would have been helpful to have some kind of benchmark to show performance improvements or cpu load reduction.

I guess this is more important for advanced features like canvas than regular browsing.

I imagine this is related with WebGL support, right?

The article says, "New APIs and markup like WebGL and 3D CSS transforms are a major motivation for this work".

That's what makes it exciting. Hardware acceleration for rendering the front page of HN: not so exciting. Hardware acceleration for 3D APIs drawing stuff in a canvas tag: very exciting.

I didn't saw libGLcore.so in `ldd chtomium-browser' output. ^_^

Reply on news.ycombinator.com