Clojure REPL为这段代码做了什么样的优化？

Question

I thought I would build a silly non-tail-recursive version of a multiplication function and see how it compares against the proper TCO one. However I noticed that both at the REPL (I am using Emacs with java -cp <classpath> clojure.main configured for inferior-lisp ) and when invoking the program from the command line some kind of optimization / memoization is apparently taking place. In fact the results are much more pronounced in the REPL.

(defn mult-silly [n m]
  (if (> n 0)
    (+ m (mult-silly (dec n) m))
    0))

(dotimes [_ 5]
  (println (time (mult-silly 5000 4)))) 

The above produces on the REPL:

user=> #'user/mult-silly
user=> "Elapsed time: 10.697919 msecs"
20000
"Elapsed time: 3.069106 msecs"
20000
"Elapsed time: 2.301474 msecs"
20000
"Elapsed time: 1.285696 msecs"
20000
"Elapsed time: 0.585541 msecs"
20000

Any idea why I see this?

Answer 1

As @MariusDanila pointed out in his comment, this is due to the JIT kicking in.

To verify this you can run java with the -Xint option, which causes it to run in interpreted-only mode, so nothing is compiled to native code (and surely no optimizations are done to that native code).

So here's what I got running java normally:

"Elapsed time: 4.175 msecs"
20000
"Elapsed time: 2.548 msecs"
20000
"Elapsed time: 7.746 msecs"
20000
"Elapsed time: 1.919 msecs"
20000
"Elapsed time: 1.72 msecs"
20000

Note that here the time actually increased for the third run. I guess this due to compilation occurring concurrently. Whereas, with -Xint :

"Elapsed time: 31.463 msecs"
20000
"Elapsed time: 30.844 msecs"
20000
"Elapsed time: 30.643 msecs"
20000
"Elapsed time: 29.972 msecs"
20000
"Elapsed time: 30.617 msecs"
20000

As you can see in the second case there is no speedup.

This is why Rule 1 for microbenchmarking is to always exclude warmup times from your measurements.