Java 8 RFR 8014319: Faster division of large integers

Tim Buktu tbuktu at hotmail.com
Wed Jul 17 20:53:30 UTC 2013


I noticed a couple of minor things.
In MutableBigInteger, I made an error in the javadoc where it says

+     * Has the same effect as {@code addend.leftShift(32*ints); add(b);}
+     * but doesn't change the value of {@code b}.

when it should be

+     * Has the same effect as {@code addend.leftShift(32*ints); add(addend);}
+     * but doesn't change the value of {@code addend}.


In BigIntegerTest, I think SIZEd should probably be lower case:

+ * throw in boundary numbers such as 0, 1, maximum SIZEd, etc.

+     * a lot of numbers that will find failure points, such as max SIZEd

Tim

 

On 12.07.2013 20:59, Brian Burkhalter wrote:
> Math Reviewers:
>
> This review request encompasses the incorporation of the Burnikel-Ziegler recursive division algorithm into BigInteger. As presented, it applies when the large integers involved are each represented by at least 50 ints. This algorithm crossover threshold is subject to change pending further experimentation.
>
> This patch is identical to the original contribution from Tim Buktu
>
> http://mail.openjdk.java.net/pipermail/core-libs-dev/2013-July/018784.html
>
> aside from some minor cosmetic javadoc updates in MutableBigInteger and the addition of the divideLarge() method and some other minor changes in BigIntegerTest.
>
> Issue:	http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8014319
> Webrev:	http://cr.openjdk.java.net/~bpb/8014319/
>
> Thanks,
>
> Brian
>
> [1] http://cr.yp.to/bib/1998/burnikel.ps
> [2] http://bioinfo.ict.ac.cn/%7Edbu/AlgorithmCourses/Lectures/Hasselstrom2003.pdf
>
>




More information about the core-libs-dev mailing list