Thomas Stuefe
@tstuefe
Followers
890
Following
3K
Media
17
Statuses
1K
@[email protected] JVM engineer at Red Hat. OpenJDK dev, Exil-Norddeutscher. Opinions are my own and do not reflect the opinions of my employer.
Mainz, Germany
Joined March 2012
Impressions from FOSDEM 25 4/4 Martin Doerr, PPC port lead, gave an interesting deep dive into the PowerPC implementation of the Foreign Function Interface, earmarked to replace JNI
0
1
4
Impressions from FOSDEM 25 3/4 Kelvin Nilsen talking about Generational Shenandoah ( https://t.co/0dv94eJv5c). Shipped as an experimental feature with the upcoming JDK 24, it shows impressive gains for most workloads.
1
1
2
Impressions from FOSDEM 25 2/4 Me talking about the future of class space. This by-product of our work on Lilliput will see us revamp class space, possibly completely removing it. That would prepare Lilliput 2 and significantly reduce maintenance costs.
2
0
2
Impressions from FOSDEM 25 1/4 @rkennke talking about Project Lilliput ( https://t.co/GEHb6nhfuI) and its possible future with Lilliput 2. I am proud to have been a part of that. We can boast significant improvements on both footprint memory bandwidth.
2
3
9
A few years ago we all met in Waldorf with @tstuefe , @volker_simonis , and Kelvin to learn about Lilliput and Shenandoah. Now both are either in or going in. Congrats! (and I need to find that picture)
And ... it's landed! π₯³π₯³ Thank you for your patience! And huge shout-out to @tstuefe and @stekarmatrik and all other contributors (too many to list them, check the PR) for all their help and support! ππ
1
2
8
We did it :-) Huge thanks to @rkennke and all the others!
And ... it's landed! π₯³π₯³ Thank you for your patience! And huge shout-out to @tstuefe and @stekarmatrik and all other contributors (too many to list them, check the PR) for all their help and support! ππ
0
1
12
The work on the JVM ecosystem always amazes me, especially how much someone has solved the problems you may be facing. In this case, I tip my hat to @shipilev @forked_franz @trustin @tstuefe, and @volker_simonis ...
3
8
42
Thanks, and good job @rkennke! Pleasure working with you :)
A big shout-out to @tstuefe and @stekarmatrikβs team without which this would not have been possible. ππ
0
0
3
Small trick to demonstrate the tighter packing of objects with a Lilliput-enabled JVM with a simple Java program: Call JVM with -XX:hashCode=4, then create a bunch of Object() and print out their ihashes. If no GCs intervene, addresses should be 8- instead of 16-byte-aligned.
1
4
8
https://t.co/XWx71E15c1 by @RedHat 's @AndrewHaley13 Bye bye type pollution issue, you won't be missed!
github.com
This PR is a redesign of subtype checking. The implementation of subtype checking in the HotSpot JVM is now twenty years old. There have been some performance-related bugs reported, and the only wa...
5
28
98
Trying to open a bug report with @jetbrains is an exercise in frustration.
0
0
1
New JEP proposal "Security Providers Filters" by our Martin Balao Alonso and Francisco Ferrari Bihurriet. https://t.co/2N7lhSLtja
0
0
0
Video of my #FOSDEM 2024 talk, βRuby on the Modern JVM: Fibers, FFI, and Moreβ is now available on the FOSDEM site below! Iβve also uploaded PDF slides which should be visible soon: https://t.co/xsBEvRcKFC
archive.fosdem.org
0
6
28
Noice, proposition by @tstuefe to have library to intercept memory allocation via the infamous LD_PRELOAD and report them in NMT. It would be useful for FFM too With the NMT JFR event that could be really nice, however those don't have stacks. https://t.co/PjGbyWwnPE
1
4
5
π #jdk22 will have a `jcmd {pid} https://t.co/yAcc3ELx5B` to show the annotated memory mapping on linux. Thanks @tstuefe for this ! Would be useful on other platforms, but that requires some work ! E.g. Is this really necessary to call another process like `vmmap` on macOS.
1
13
30
We can now limit the memory used by the JIT (C1, C2). That allows us to cut off those rare spikes that can cause lasting OS-side memory retention. https://t.co/ouenDU3cCh
github.com
This RFE introduces a new compile command to limit memory usage per compilation. It complements and is based upon the JIT memory statistic introduced with https://bugs.openjdk.org/browse/JDK-831768...
0
9
42