为什么这两个反向循环之间存在性能差异？

Question

Benchmarks

Note: To run just the following two tests on that benchmark suite, click on Basic reverse for loop and Falsy reverse for loop .

The ops/sec listed below are on Chrome 32, Win 7 64-bit.

---

This loop

for (var i=a.length - 1; i >= 0; i--) { }  // 1,161,089 ops/sec

is much faster than

for (var i=a.length; i--;) { }  // 870,837 ops/sec

on Chrome 32, Firefox 27, and Opera 12 & 19. They're about equal in IE (5 - 11) and the falsy loop is actually faster in Safari Windows (5.1.7). This doesn't seem to have anything to do with the loop conditional being falsy - there's another benchmark in that suite for that comparison.

The "basic" loop is about 33% faster than the "falsy" loop on Chrome 32, Win 7 64-bit. Why?

Answer 1

That second loop: for (var i=a.length; i--;) { } iterates until i-- returns false, since that part is written in the condition part of the loop. i-- will return false when i equals 0, so the functionality of the two loops is the same.

However, the second loop causes the additional task of casting each of the values between a.length and 0 to a Boolean, which is more time consuming than integer comparison.