Website load times on ARM Cortex A8-powered mobile devices
While assessing my mobile gadget collection the other day, I realized that four of my devices are powered by CPUs with ARM Cortex A8 cores. For nearly eight months, my non-Android Archos 5 IMT (600MHz TI OMAP 3430) had bragging rights as the only Cortex A8 handheld in the house, so I didn't notice that it lost its prestigious title when the iPhone 3GS (600MHz Samsung S5PC100) waltzed in this past summer, the Sharp NetWalker (800MHz Freescale i.MX515) moved in last month, and the Archos 5 Android tablet (800MHz TI OMAP 3440) arrived a few weeks ago.
I've been very preoccupied with matters unrelated to portable gadgets for most of the year (and I usually don't pay much attention to internal components anyway), so the realization that a single Cortex A8 device turned into four Cortex A8 devices when I wasn't looking was a little alarming. The only thing to do, then, was to pit the units against one another for a round of website load time tests.
Seeing that the Archos 5 Android loaded web pages faster than the Archos 5 IMT in last week's browser battle already confirmed to me that all Cortex A8 cores are not created equally, but I couldn't resist putting another test together just for kicks.
I know that a device's processor isn't the only factor in how quickly a website loads (browser, ad servers, RAM, and other potential bottlenecks all come into play as well), but the Cortex A8 core is the only common denominator when looking at the iPhone 3GS, Archos tablets, and NetWalker as a group.
As usual:
- This was not a scientific study or professional lab test.
- All accessed pages were full versions (not mobile versions) of the website.
- Load times were measured from the same location connected to the same WiFi network from the click of the OK/Go key to the complete page load according to the progress bar.
- All devices are running the most recent firmware as of today.
- Each website was loaded three times after an initial cache clearing and the results were averaged.
- Load times vary by location, time of day, ad servers, content, etc., so your results will not be identical to mine.
Some have argued that timing how long a page takes to load completely isn't an accurate measurement, as portions of the site are often readable/usable before then, but I think it's one of the most objective methods. Timing how long a site takes to partially load is much more subjective, as someone who only wants to view the second story on a website will often be able to "use" the page much sooner than someone who wants access to the tenth story.
| iPhone 3GS Safari |
NetWalker Firefox |
Archos 5 IMT Opera 9 |
Archos 5 Android WebKit |
|
| Amazon | 8.7 seconds | 12.1 seconds | 10.4 seconds | 8.6 seconds |
| CNET | 8.9 seconds | 10.6 seconds | 20.0 seconds | 8.7 seconds |
| Engadget | 12.4 seconds | 15.3 seconds | 12.7 seconds | 10.9 seconds |
| Flickr | 5.5 seconds | 5.1 seconds | 4.4 seconds | 5.9 seconds |
| Google News | 5.5 seconds | 5.5 seconds | 9.1 seconds | 6.3 seconds |
| MySpace | 7.1 seconds | 8.3 seconds | 8.8 seconds | 6.2 seconds |
| NY Times | 17.0 seconds | 18.2 seconds | 22.5 seconds | 16.4 seconds |
| Pocketables |
12.6 seconds | 13.1 seconds | 12.5 seconds | 10.6 seconds |
| Twitter |
7.5 seconds | 5.4 seconds | 6.3 seconds | 6.9 seconds |
| YouTube |
10.7 seconds | 9.1 seconds | 9.7 seconds | 9.7 seconds |
| AVERAGE | 9.6 seconds | 10.3 seconds | 11.6 seconds | 9.0 seconds |
The differences between the average page load times are not extraordinary, as only 2.6 seconds separates the fastest device from the slowest, and there's much more at work here than just different processors (e.g., operating systems, web browsers, etc.), but I still think the results are pretty interesting.
