Apple has released another update of its Safari browser for Windows, and claims in usual hyperbolic Cupertino fashion that it is “the fastest, easiest-to-use web browser in the world.” Mozilla is on beta 4 of the next version of Firefox, and Microsoft released IE7 not that long ago (and is beginning to test IE8). So I decided to do a few quick tests to see how much Safari has improved. When it was originally released, I kept it installed for about 15 minutes before returning to FF. For purposes of the following, I created a new FF profile with no extensions, but kept everything else as a stock install. The test PC is a Compaq Presario A900 notebook (1.6 Ghz dual-core Pentium, 2 GB RAM, Vista Home Premium SP1), which is squarely middle-of-the-road these days.
For the first test, I loaded the default home page of each browser (MSN for IE, Firefox Start for Firefox and Apple Start for Safari) and used the Vista Sphere Timer gadget to time the startup time. I did this three times and averaged the results:
IE7 | Firefox 3.0 b4 | Safari 3.1 |
7.15 sec | 5.70 sec | 6.63 sec |
Firefox has an advantage in this test, as the Google-driven start page is sparse and relatively graphic-free. While I’d argue that this is relevant as indicative of design philosophy, nevertheless I set the default home page to be about:blank in all three browsers and ran the test again:
IE7 | Firefox | Safari |
4.04 sec | 3.30 sec | 2.64 sec |
Safari is noticeably zippier when not asked to load the Apple start page. How many users ever change the default start page, however? In any event, the differences here are minor. Apple also claims superior HTML and JavaScript rendering speed for its new browser, based on tests using iBench 5.0, a test suite developed in 2003 by PC Magazine and VeriTest. It has been criticized, however, for giving Safari an advantage because Safari reports that a page is loaded before calculating layout of the page. Other tests show different results. I decided to use the JS test at Celtic Kane, running each browser through the test ten times:
IE7 | Firefox | Safari |
1336.3 ms | 676.7 ms | 394.6 ms |
Test results are only as good as the test, but this test tends to support Apple’s claims regarding JavaScript speed. I have not, however, noticed much of a real-world difference on AJAX-heavy sites like Gmail. Finally, I opened four tabs in each browser to check memory use. In this case, two different Gmail accounts, Google Reader and my blog Dashboard at wordpress.com. It appears Mozilla’s efforts are paying off:
IE7 | Firefox | Safari |
189.96 MB | 63.64 MB | 131.62 MB |
Benchmarks don’t tell the whole story, of course, and each browser has additional advantages. Safari has SnapBack, resizable text areas and proprietary color and font management. Firefox has an open and extendible structure, which allows users to add virtually any conceivable functionality via extensions. IE has ActiveX (which is as much curse as blessing, of course, and achievable in the other browsers with some tweaking) and a shrinking but still sizable library of sites that work better (and sometimes only) in IE. The latter also comes preinstalled on the dominant OS, obviously. I plan to use each browser extensively over the next week or so and cover features and real-world performance later.