I think the 1.2 times faster number is from the linuxdocs PHP HOW-TO. That How-to is no longer available on the linux tldp site. I found it on various mirrors, the 1.2 times figure is measuring some kind of raw page outputs. Unfortunately all the links to benchmark specifics in the how-to copies were dead.
Specific benchmarks aren't that useful in the real world anyway, imho. One of ffb's links in this thread was a speed comparison of some kind of fractal math routine over many repetitions. Since I mostly do database small business sites that's not a speed comparison that means much in my world To me, overall throughput and performance is much more important than some benchmark number from some specific controlled non-real-world test.
I do some development work with an open-source ASP bbs, I've evaluated many, many asp bbs's, and I've recently been looking at PHP bbs's. There is very little difference in speed between the various packages I've played with in a low-use environment, and I'm pretty convinced the biggest speed differential between packages is the dataase design, not PHP or ASP as the web server.
$0000000000.02