The problem is, benchmarks are never a fair game. It's the nature of the benchmark as a genre. You can always bias a benchmark towards some code path to show that you're superior.

Also, there are other factors to consider:

- Some of the benchmarks are "lower is better", so reading Y axis is important.

- Some results are very close (e.g. speedometer), but the zoom makes difference bigger, so reading the Y numbers again is important.

So, Firefox beats Chrome on WebAudio, StyleBench, AssortedDOM. However, this is still "benchmarks", The real world performance is very, very close.

The bigger picture is, when you look at longer histories, the performance is still being tuned and improved. So, Firefox people are not sitting on what they have.

Lastly, Firefox is way more sensitive to DNS response time when compared to Chrome, and a crowded site makes tons of requests. A fast DNS makes a ton of difference, which is way overlooked.

I used to run a DNSMasq instance when my ISP DNS was very slow. Now, my routers have their own tuned DNSMasq instances, so DNS is instant, so Firefox is as well.