Since I don’t have all the time in the world, I haven’t been benchmarking the newer builds I’ve posted lately. In particular, I’ve never benchmarked the 64-bit Firefox 3.7a1 against the 32-bit nightly. A kind soul pointed out that my builds yielded no speed gains and are inferior to the official Mozilla nightlies. Having done numerous builds of 64-bit WebKit (pre-WWDC), and testing against the 32-bit WebKit and finding no gains, I didn’t doubt Firefox suffered the same fate.
But it seems faster. Everything seems faster. New iPhone OS releases always seem faster, new point releases of OS X seem faster, new stuff just seems faster. So I decided to spend most of my day trying to quantify this seemingly faster build.
Here’s what I got.
I ran this three times for each browser and took the average. The beetle build wins by 2%!!
V8 Benchmark Suite – Version 5
Ran this three times and took the average. The beetle build wins by 5%!!
Some cheez generic browser benchmark site. Never heard of it. Ran the test three times and took the average. Firefox nightly wins by 3%!!
Here is where the seeming speed gets put to the test. Loaded a cross section of 25 sites in 25 tabs. Used the ol’ iPhone stopwatch, ran three times from a warm launch and took the average.
Well, look at that. The beetle build whomps Firefox by over 30%!!
Yes, the nightlies might be better. I’m pretty sure they’re more stable, I’m positive they’ll at least load Flash, and they’re probably even better optimized for your machine. My builds are optimized for myself, more tuned for 10.6.x and 64-bit so I’m going after the bleeding edge.