Consistent way to test browser speed?

[ervus]

I've been looking for a consistent way to test different TFF settings and hardware setups for speed. The threads here have a lot of suggestions for settings, but when I try to toggle things or make changes it's difficult to measure a difference. It seems most web sites now are a rats' nest of different servers and scripts. This page I'm typing in has stuff from at least 7 servers. Just loading the same page two different times with the exact same setup will take different times. There are a variety of hardware setups that I would also like to test. How much would a SSD help browsing speed. What about 1.5GB of RAM vs 2? etc.

So what is a consistent way to test browser speed?

What if I save a local copy of a couple different web pages, then open a few tabs and load those pages, then save/exit the browser session. Then time how long it takes to open TFF and reload the session from disk. Suggestions?

 

[eyoungren]

It's really hard to answer this question because some of this is based on your…settings.

Do you use cache? If you do, how are you using it? You can offload cache to a different drive. I used to have cache set to a virtual disk, which meant cache was always loaded in memory - so access was instantaneous. Do you preload links or images? Are you painting content as soon as it arrives or waiting for the page to render completely before displaying? Are you using a stable connection and what kind of speeds go you generally get? How about your router and modem? Are you using WiFi or Ethernet?

You can get seriously wrapped around the axle on all of that. Here's a better idea. Using a stopwatch, load a page. Start the time when your CPU spikes. Stop the time when your CPU returns to it's normal idle level.

The time it takes to go from the max back to idle speed is the time that the CPU has spent processing and rendering the page. That's what's important. All your setting changes actually affect that process so that's what you should be measuring.

 

[ervus]

So here was a simple test. I pulled up activity monitor and then hit reload on the 'PowerPC Macs' forum page:

The difference is the DFS setting. That at least shows the simple effect of CPU speed with everything else unchanged in a back-to-back comparison.

 

[eyoungren]

Are you using the rich text editor on MacRumors? You can tell if you are by looking in the post editing box. If the icons there (bold, italic, etc) are not grayed out you'll want to turn the editor off. Press the hamburger icon on the far right (by the preview button) and then press the two brackets button. That turns off (and on) the editor.

When the editor is on it consumers CPU power because it's javascript that is running continuously in the background. With it off, your CPU won't be hammered.

Of course, you need to learn to hand code, such as open bracket b closed bracket for bold open bracket forward slash b closed bracket, but that comes easier once you learn it.

 

[ervus]

That was just a simple example. What I'd like to do is come up with some sort of browser test that I can use on a different computer, or on a different day, without internet, to experiment with hardware/software setups.

For instance, if I want to try and answer a question like: how does an AGP with a single 1.8GHz CPU compare to a Quicksilver with a dual 1GHz CPU? There are a number of hardware differences between these two, so how can I minimize the software differences, or network differences, etc.?

In TFF I can 'Save Page As...' and under that is an option for 'Web Page, complete' so I'm guessing this would save all the scripts and other junk in a way that would let me consistently time how long it takes to open or refresh it?