11 comments

[ 4.3 ms ] story [ 14.0 ms ] thread
Perfect example of data representation manipulation:

- take a benchmark created by producers of A and show results for products A-Z

- present the data which does not show a percentage (or "add up" to anything specific), on a pie graph

- take out an arbitrary element which resizes the proportions, to "make the comparison even more striking"

Indeed. Are there really people who look at pie charts and don't read it as representing 100% of something?

What exactly does each slice of that pie represent? X% of performance?

I sure can't wait till my browser claims 100% of performance!

Don't forget to use a 3D pie chart with A closer than Z so that A will appear to have a larger share than it actually has.
Please, please, pretty please: If you use charts, choose a sensible type of chart to illustrate your point.

A pie chart is simply wrong for representing a comparison of benchmark results. A bar chart would be the correct type, in this case.

Agreed.

As if that weren't bad enough, the charts nor the article contain information about what's been tested. Is it the speed? The memory-usage? Something else?

The results themselves are pretty interesting though. Only thing I feel left out is Opera.

The pie charts are really really irritating.

From the original article (which also mentions Sunspider benchmark.. i always thought it may be strange to use the V8 benchmark as hard evidence, as it comes from Google/Chrome):

The SunSpider test took just 345.6 ms on our quad-core test system, compared to 387.8 ms of Chrome 5.0.342.8. Chrome 4.x scored 483.3 ms on this system. Now keep in mind that Chrome 5 is still in beta, but it is already 30% faster than Chrome 4, which remains the fastest (stable) JavaScript browser currently available. In comparison, the current Firefox 3.6.4 scored 876 ms, Safari 4.04 530 ms, IE8 4979ms and the IE9 Platform Preview 783.6 ms. In that view, Chrome 5 is more than twice as fast as IE9 PP.

It's good to see such big improvements in IE9.. now we only need to get rid of IE6-8 :(

What's with the blatant ignoring of Opera here? It shows up in the article's source link's market share, but in none of the performance comparisons. This line is particularly egregious:

>Chrome 4, which remains the fastest (stable) JavaScript browser currently available.

Chrome does beat Opera on V8 (which isn't surprising, since Google created V8), but on Sunspider, well...

>The SunSpider test took just 345.6 ms on our quad-core test system, compared to 387.8 ms of Chrome 5.0.342.8.

And here are my results in Opera 10.53, using a machine that's over a year old (AMD Phenom II X4 940, factory settings):

  Total:                  302.8ms +/- 2.1%

Edit: Removed full results because it made my comment too long, but if anyone's actually curious for more detail I'm happy to run it again.
you really can't compare your timing results with theirs, as you don't have the same machine. Run the same test on your machine with Chrome 4 and 5 and then compare those.
Chrome 5 was the clear performance winner. "In Google’s own V8 benchmark ..."

Google browser and Google benchmark? Can we get a benchmark from an unbiased source?

You often have to wonder if browsers only get better benchmark times because the benchmark times are targeted. I.e., the brain gets faster at doing the same thing over and over again, so the programmers are tuned to getting a faster benchmark time, and optimize accordingly.
If anyone on the Chrome team is reading, the omnibar is awesome, but please give us the option to also have a multisearch box like FF and IE.