Re: GC Help

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re: GC Help

Hohensee, Paul
Also, the server compiler makes every effort to remove zeroing fields if the constructor initializes them. Much/most zeroing ends up being elided.

Paul

On 10/24/17, 2:35 AM, "hotspot-gc-dev on behalf of Andrew Haley" <[hidden email] on behalf of [hidden email]> wrote:

    On 24/10/17 07:24, amith pawar wrote:
    > Is it good to have bulk zeroing feature/pass in GC where once sweeping is
    > done then do bulk zero ? If this is done then during allocation step
    > zeroing is not required as it is done in GC itself.
    >
    > Not sure how much time GC will take for this.
    >
    > I may not be the first person to think this way and thought to ask here.
    > Does JAVA spec allows this (such pass in GC)? if so any pointers on that.
    > It will help me to understand more.
   
    It's called -XX:+ZeroTLAB.  See also ReduceBulkZeroing.
   
    It may or may not be effective: zeroing at allocation time is fairly
    cache friendly, whereas zeroing when allocating a new TLAB isn't going
    to be.  Some architectures have cache-missing bulk zeroing, which
    might help.
   
    --
    Andrew Haley
    Java Platform Lead Engineer
    Red Hat UK Ltd. <https://www.redhat.com>
    EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671