Review: Google Chrome beta released for Windows

Shortly after 12:00 PM PDT today, Google released the beta version of its new web browser, Google Chrome. Currently it’s only available for Windows Vista and XP, and only in 32-bit. I downloaded it and took it for a test drive.

Whenever I install software on Windows, I always customize the installation to eliminate the desktop icon, if that option is provided. I wasn’t disappointed:

Note that Chrome can import your settings from Firefox and others. I thought it was interesting that “Make Google Chrome the default browser” was not checked by default.

Here’s the default start page, which includes your most visited sites and recent bookmarks:

As you can see, Chrome uses tabs like other browsers, but it doesn’t use tabs like other browsers. According to the story, each tab is a separate process (which I verified in Task Manager). You’ll note also that each tab gets its own address bar. I was relieved to find that Ctrl-T opens a new tab, just like in Firefox.

I found it odd that Chrome overrides my Windows theme and chooses its own rendering for the containing frame (I’m using the Vista Basic theme, which should have colored the frame in a graduated gray-blue. It’s not annoying, but it’s a “does not play well with others” feature.

There is no separate “search” box in Chrome. Rather, you use the address bar, which Google calls the “Omnibox” (maybe for omniscient). For instance, when I type “google chrome” in the Omnibox, I get:

The suggested results include not only recently visited sites that mention “google chrome”, but also popular Google search results for “google chrome” and an explicit “Search Google for chrome”. No more remembering which box to type into.

First impressions: fast. Much faster than Firefox. There are several possible reasons. First, keeping each tab in its own process allows for a cleaner memory management model.

Second, Google has kept this product very simple so far. Alas, that means there is no facility for extensions short of downloading the open source and adding what you need. Simple is also why they chose the Apple WebKit rendering engine.

Finally, they’ve created their own JavaScript engine called V8, which provides the following speed improvements:

  1. Multi-threading. I’m not sure what challenges that may pose for existing single-threaded code, or how to make use of multi-threading explicitly.
  2. Dynamic compilation to native code. Zoom.
  3. Hidden class transitions: optimizing for repetitive composition of similar classes.
  4. “Precise” garbage collection. They keep better track of pointers to objects, and release them as soon as they are no longer referenced.

I ran Mozilla’s own benchmark test for JavaScript at dromeao, comparing various browsers. Depending on the operation, Chrome could be roughly 2 to 20 times faster than Firefox or Safari (Internet Explorer 7 kept timing out on me, so you can assume it’s much worse). You can peruse the hard numbers if you’re curious for Google Chrome beta, Firefox 3.0.1, and Safari 3.1.2, but here’s a visual (shorter bars are faster):

Since Chrome boasts of its memory management model, I decided to load up the same four sites in both Chrome and Firefox to compare. Firefox’s one process consumed 84 MB. Chrome’s six processes totaled 91.5 MB. As I closed the tabs down to one, however, Firefox dropped only down to 75 MB, while Chrome’s three processes totaled 49 MB. Reloading the same three pages I just unloaded bumps Firefox up to 88 MB, but Chrome only came back up to 89 MB total. Firefox users are all too familiar with Firefox’s memory creep syndrome to fail to appreciate this difference.

The options dialogs are pretty straightforward as what you would expect to find, but I was impressed by the built-in developer features:

View source automatically comes up in a new tab instead of a separate window (nice). I haven’t figured out how to use the JavaScript debugger yet, but the JavaScript console appears to have some nice features:

Another feature that deserves mention is “incognito mode” (New incognito window, or Ctrl+Shift+N), which opens a new browser that will save none of your personal information — aka “porn mode”. As they warn in the Learn more link, this will not prevent web sites you visit from gathering information from your session, however.

Overall I’m really impressed with this browser. The one thing I’d miss most if I switched would be my beloved Firefox extensions. But the speed difference is impressive.

On the other hand, Shelley Powers expresses concern over how well Chrome will support emerging standards, like HTML5 and the new ECMAScript standards. She also questions Google’s motives for introducing yet another combatant into the browser wars. Obviously Google wants to increase its influence on the web, and perhaps building their own client may add leverage to their applications. Will we come to a point where Google can say, “best viewed with Chrome” and let other browsers fend for themselves? Will future standards be dictated by what Google wants? At least the project is open source, so other browsers can learn from Chrome if they so choose.

But just how revolutionary is this browser, anyway? In an email to me before Chrome was released, Shelley quipped: “Hard to say, all we’ve seen is a comic book. Other comic books bring us people who can fly and spin webs–what Google promises with Chrome could be just as far fetched and fanciful.”

If you’re looking for a grand new design for the client side of the web, then you’d be expecting too much. What Chrome provides is more of the same paradigm, but more efficient and arguably more user-friendly.

What do you think of Google Chrome? Have you download it yet?