So your pages weights 170KBytes.

Dialup connection is 56Kbit/second = ~5KB/second, so it will take 170/5=34+lag
= ~40 seconds to fully load this page.

If you compare this to broadband 1Mbit/second = 128Kbyte/second it will be 170/128=1.3+lag (lower for fast connection)
= ~1.5 seconds to fully load same page on such connection.

This is simplification of course. There are many other factors that contribute to load speed so it’s theoretical, while real page size is factual.

Difference in size is definitely because of compression. Look at “Components” tab in YSlow – SIZE and GZIP columns (expand types to see gzip values).

Size is what real size of component is. gzip is compressed size that is really transferred if browser supports compression (complicated topic, but modern browsers mostly do).

Text resources like HTML or JavaScript code are highly compressible. So while this page is large itself and has some large scripts – it gets much smaller when compressed for transfer.

Difference in load speed – as above, connection issues. Pingdom server has faster connection so faster actual load speed. Such load speed is naturally inaccurate as metric.

For simple overview I suggest you use weight graphs (Statistics tab in YSlow). They show which parts of site is heavy and are easy to get.

If you need range of load speeds just take site size and calculate approximate times for different connections manually, then increase a bit to account for latency of multiply components.

This would be one of the page I am currently looking at:
http://bit.ly/3xR7EQ

Got a size of 207.7KB in Yslow, 123KB with Page Speed, 350KB with Pingdom
Then I got between 1.8&2.4sec loadtime in Pingdom, 5.93sec with Yslow, 4.2sec with LORI (both local tools launched at the same time)

I’d like to give a max. load time & size to my tech collegues and am therefore benchmarking a few sites, but really get all types of numbers…

PS By similar, I meant same page, language problem, sorry :)

What do you mean similar pages? To get close results it makes sense to test single and exactly same page.

No idea on number of images. My only guess is that some tools (I think pingdom is one of them) may at times include images referenced, but not actually used in CSS.

Time to load naturally varies with connection speed and quality. For YSlow and PageSpeed this would be your own Internet connection, for online tools this would be their server’s.

Size of elements (mostly HTML, CSS and JS) can vary because of compression used. Not all tools notice or support compression and some only report uncompressed sizes.

If you provide link you are attempting to benchmark I will try to take a look. :)

Glad you like it. :) YSlow is awesome but doesn’t give clear picture of download process. Actually Firebug has such graph, but bit boring and hard to interpret.