Hi all!
I would like some help in trying out some experiments in quantifying user interface "snappiness" and would like your help. No software required to install!

My goal is to test if using the BFS scheduler (http://talk.maemo.org/showthread.php?t=58780) results in a user interface with less latency.

Needed:
- A Nokia N900
- What kernel you are using. Stock or powersave or other? Only considering stock kernels would make things easier.
- Use of the email client that is configured to check an account via IMAP. Preferably a gmail account with a few thousand messages.
- An internet connection via WiFi that is fast enough to make the N900 the limiting factor in browsing the web.
- The stock transition settings.
- A stopwatch.
- The stock CSS applied when visiting nytimes.com. No flashblock or adblock. Or if you do have flashblock or adblock at least tell me ;-)

The idea is to give the N900 a reasonable workload and then measure the user interface latency via a UI heavy operation.

procedure:
1. Initialize the N900 to the following state: The email application is open, at the inbox screen, and not checking email. The web browser launcher is open and microb is started. (I think you have had to at least visited a single web page since boot to start microb?)
2. enter, but do not go to, www.nytimes.com in the launcher URL bar.
3. Go to nytimes.com and note start time
4. Switch to email via ctrl-del and then tapping the email image.
5. Select "Send & Receive" from menu
6. Repeat the following procedure *at least* 15 times:

a. switch to the nytimes browser via ctrl-del then tapping the browser image.
b. switch to the email application via ctrl-del then tapping the email image.

7. note the time

Finally: Post the difference between end time and start time together with how many times you repeated step 6.

Lemme know if you can think of any refinements to this test. I can think of a few problems but in general I think it should work OK.

Won't work for me, am on the other side of the globe linked through two
geosynchronous satellites ported through tunnel buried in a hostile ISP.
I am looking to run hardinfo benchmarks if I can get it compiled...