Custom Query (1030 matches)

The following log probably describes the bug fully. I'm not sure how to assign the priority, but this seems major to me. It occurs in 16119 but not in 15915.

dunnottar:~% uname -a
Linux dunnottar 3.2.0-64-generic #97-Ubuntu SMP Wed Jun 4 22:04:21 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux
dunnottar:~% ccl
Welcome to Clozure Common Lisp Version 1.10-dev-r16119M-trunk  (LinuxX8664)!
? (defun foo (ar)
    (* (aref (the (simple-array fixnum (*)) ar)
	     0)
       (the (signed-byte 61) 1)))
> Error: The value NIL is not of the expected type INTEGER.
> While executing: INTEGER-LENGTH, in process listener(1).
> Type :POP to abort, :R for a list of available restarts.
> Type :? for other options.
1 > 

The log below shows an incorrect evaluation result.

I've marked this as "critical", but perhaps it should be "blocker". ACL2 relies on such multiplication being correct for its own checksum computations. We are hoping to release a new version of ACL2 this week.

Welcome to Clozure Common Lisp Version 1.10-dev-r16148M-trunk  (LinuxX8664)!
? (defun foo (u v)
    (declare (type fixnum u v))
    (* u v))
FOO
? most-positive-fixnum
1152921504606846975
? (foo 100000000000 100000000000)
-441130959790669824
? 

A colleague noticed that the following function can run more slowly with a suitable type declaration. I've reproduced his results on Linux:

Welcome to Clozure Common Lisp Version 1.12-dev-r16729M-trunk (LinuxX8664)!

Here are the commands:

(declaim (OPTIMIZE (COMPILATION-SPEED 0) (DEBUG 0) (SPEED 3) (SPACE 0) (SAFETY 0)))
(defun fibonacci (n)
  (declare (type (integer 0 *) n))
  (if (= n 0)
      0
    (if (= n 1)
        1
      (+ (fibonacci (- n 1)) (fibonacci (- n 2))))))

Then:

? (time (fibonacci 38))
(FIBONACCI 38)
took 912,371 microseconds (0.912371 seconds) to run.
During that period, and with 8 available CPU cores,
     912,812 microseconds (0.912812 seconds) were spent in user mode
         804 microseconds (0.000804 seconds) were spent in system mode
 1 minor page faults, 0 major page faults, 0 swaps.
39088169
? 

But here we see the time cut in more than half when we remove the type declaration:

? (time (fibonacci 38))
(FIBONACCI 38)
took 383,980 microseconds (0.383980 seconds) to run.
During that period, and with 8 available CPU cores,
     382,068 microseconds (0.382068 seconds) were spent in user mode
       2,433 microseconds (0.002433 seconds) were spent in system mode
 1 minor page faults, 0 major page faults, 0 swaps.
39088169
?