RFR: 8357531: The `SegmentBulkOperations::fill` method can be improved using overlaps [v5]

[Per-Ake Minborg]

Hi Archie!

The C2 compiler can use automatic vectorization [1] to speed up such operations. You can watch my explanation of the concept in some of my recent presentations, e.g, [2]

Best, Per
YouTube<https://youtu.be/rXv2-lN5Xgk?t=1518>
Share your videos with friends, family, and the world
youtu.be

Automatic vectorization - Wikipedia<https://en.wikipedia.org/wiki/Automatic_vectorization>
Automatic vectorization, in parallel computing, is a special case of automatic parallelization, where a computer program is converted from a scalar implementation, which processes a single pair of operands at a time, to a vector implementation, which processes one operation on multiple pairs of operands at once. For example, modern conventional computers, including specialized supercomputers ...
en.wikipedia.org
[1] https://en.wikipedia.org/wiki/Automatic_vectorization
[2] https://youtu.be/rXv2-lN5Xgk?t=1518
________________________________
From: core-libs-dev <core-libs-dev-retn at openjdk.org> on behalf of Archie Cobbs <archie.cobbs at gmail.com>
Sent: Thursday, May 22, 2025 10:58 PM
To: John R Rose <jrose at openjdk.org>
Cc: core-libs-dev at openjdk.org <core-libs-dev at openjdk.org>
Subject: Re: RFR: 8357531: The `SegmentBulkOperations::fill` method can be improved using overlaps [v5]

On Thu, May 22, 2025 at 3:31 PM John R Rose <jrose at openjdk.org<mailto:jrose at openjdk.org>> wrote:
>   Update benchmark to reflect new fill method

Related discussion at the hardware level:

https://github.com/openjdk/jdk/pull/25147#issuecomment-2902463076

This discussion spurred me to ask a dumb question. Apologies in advance, just trying to learn here...

If I do this:

import java.util.Arrays;
public class ArrayFiller {
    public static void main(String[] args) {
        while (true) {
            final byte[] array = new byte[1000000];
            Arrays.fill(array, (byte)0x42);
        }
    }
}

Will C2 compile Arrays.fill()<https://github.com/openjdk/jdk/blob/139a05d05959a84541a29dfae6151f92ce579ae6/src/java.base/share/classes/java/util/Arrays.java#L3275-L3308> into something that is more efficient than a byte-at-a-time loop like what appears in the source code?

Thanks,
-Archie

--
Archie L. Cobbs
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/core-libs-dev/attachments/20250523/bcb1cb16/attachment-0001.htm>