rc3.org Strong opinions weakly held

Web site CPU usage

Brent Simmons says that developers should profile their Web pages to see how they affect the CPU usage on end user computers. My wife uses an iMac G5, and I can tell which Web sites she’s using based on how loud the fan is. There are plenty of Web sites that keep the fan running all the time, even if she walks away from the computer. This is an area of Web development that has been ignored but should not be.


8 Comments

I’ve got a Nokia N810, and lack of forethought along in this vein hits me especially hard. When I turned off Flash, it was revelatory, but some sites download some serious JavaScript.

What would be nice is for G-mail to recognize that I want “old-school” on some devices, and “new-school” on other, so I don’t keep switching back and forth.

Posted by Jason! on 12 December 2008 @ 8am

The proliferation of Javascript-based UI toolkits (Yahoo, GWT, JQuery, etc)–which is, in my opinion, a very good trend in general–probably exacerbates this problem.

Posted by Stan Taylor on 12 December 2008 @ 11am

turning off Flash helps, but some sites do something with their Flash windows that means you can’t load them manually, meaning you’re wrecked.

I would love sites to take load into account. advertisers will make that hard, but not impossible…

Posted by acm on 12 December 2008 @ 12pm

The better JavaScript libraries are heavily tested for load time and performance impact, but that of course doesn’t prevent people from using them in dumb ways.

Posted by Rafe on 12 December 2008 @ 1pm

You really see this with a low-power computer like an Atom-powered netbook. CPU usage near max, fan on high, (and battery life shortend), sluggish performance, just to display blinky ads and who knows what else but not some useful function you really wanted and which reasonably warrants high CPU usage.

Posted by Beige on 12 December 2008 @ 9pm

I’ve found this to be a surprisingly noticeable advantage for the newer JavaScript engines: for many sites, this makes a very noticeable difference because it’s the difference between response time being over the threshold where things appear to happen as soon as you click. This is particularly glaring if you’re unfortunate enough to be using IE (even IE8) and are suddenly forced back into the click-wait-click style of interaction.

Posted by Chris Adams on 13 December 2008 @ 4pm

Boy, no kidding. I have run a menubar CPU monitor for years (either MenuMeters on <OS X 10.4, or iStatMenus under Leopard), and it’s appalling how some sites will pin the CPU – usually over something incredibly stupid, like WRAL’s fancy new weather report page, which took a perfectly functional static image with text underneath containing the week’s weather forecast, and replaced it with a Flash-based abortion that you have to scroll to see the forecast, and which seems to have been done solely for the purpose of allowing animated sunshine, rain, and snow – and which has the effect of raising my CPU temps from the typical 58 degrees or so to the 70 degree mark in seconds, and uses 70% of the CPU – for information that never changes after page load. Wish this practice of profiling were more widespread.

Posted by Steven Champeon on 18 December 2008 @ 11am

I used to hear, “We need our site to be dynamic on every page build.”

Why?

“Well, because.”

I’ve never understood that.

Posted by Nathan L. Walls on 20 December 2008 @ 2pm

Leave a Comment