The result is faster-loading pages, more powerful Web applications, and another round in the browser performance competition with Microsoft's Internet Explorer, Mozilla's Firefox, Apple's Safari, and Opera.
"Crankshaft uses adaptive compilation to improve both start-up time and peak performance. The idea is to heavily optimize code that is frequently executed and not waste time optimizing code that is not," Google programmers Kevin Millikin and Florian Schneider said yesterday in a company blog post.
All these areas and more are getting ever more attention. And if it wasn't clear what's at stake, look no further than Google's Chrome OS and Chrome Web Store. The first is a browser-based operating system that runs Web apps only; the second is a distribution mechanism to find and buy those apps.
There are plenty of uncertainties about how well Google will succeed in its ambition to transform the Web into a foundation for applications, not just static Web sites. But there are some things that aren't so unclear: more and more of people's work and personal life is being spent doing things within a browser. That trend is enabled by better performance and, at the same time, encourages even more advances.
The programmers specifically pointed to improvements in Gmail loading times, which I've found excruciatingly slow in recent months. However, my not-terribly-reliable stopwatch tests showed Crankshaft actually slower with that site: 2.4 seconds to load on an average of five runs loading Gmail on Chrome Canary 10.0.603.3 compared with 2.1 seconds for the newest stable version of Chrome, Chrome 8.0.552.215. Given the variability in the results (less than 2 seconds to more than 3), though, I wouldn't read too much into that result.
Here, Crankshift definitely shows a difference, except on the SunSpider test whose influence has waned as browser makers' advancements have rendered it out of date. Bear in mind, though, that this was a test just on a single machine, a quad-core Dell Studio XPS 16 with 6GB of memory and that other machines will produce different results.
Browser benchmarks are a thorny issue. It's always tough to represent the full breadth of computing challenges in a single convenient test, and there's always the risk that engineers will design products for good benchmark scores even when the approach has little or no bearing on real-world work. Indeed,with IE9.
, on the verge of 10 percent of browser usage on the Web today for third place after IE and Firefox. It took years and a somewhat subversive effort to convince Google Chief Executive Eric Schmidt that the company should release a browser, but it's clearly a force to be reckoned with on the Net.