Benchmarks such as the PHP / GCC / ICC Benchmark I posted quite a while ago on this blog are synthetic. They test "raw bytecode execution" speed that cannot be translated into real-world situations per se. Flickr recently migrated from PHP 4 to PHP 5. And here are their real world numbers.
The graph shows a dramatic drop in processing needed (CPU usage) at the moment they made the switch over from PHP4 to PHP5.The system processing usage stayed about the same (system processes like logging and memory handling) but the user CPU usage (like what the web server would use) saw an immediate change by almost twenty percent.