| WEB | http://jpbempel.github.io |
| Blue Sky | @jpbempel.bsky.social |
IntelliJ IDEA 2023.1 EAP contains a nice feature: it helps to complete -XX VM options when you are editing the run configuration. Just type -XX: and start typing substring of the option you need.
🐦🔗: https://twitter.com/tagir_valeev/status/1620354382620672002
For the last half year I developed an IntelliJ plugin based on Firefox Profiler, to view JFR files directly in IntelliJ and to profile applications using async-profiler and JFR: I just released the early beta of Java JFR Profiler
https://mostlynerdless.de/blog/2023/01/31/firefox-profiler-beyond-the-web/
🐦🔗: https://twitter.com/parttimen3rd/status/1620373834384936961
Looks at TR 3970X core-to-core latency figures in oscillating awe and disgust. This is a part of "Yea, run 2-thread producer-consumer benchmarks on a large and busy machine, what could possibly go wrong".
Another example of what RyuJIT is capable to fold in .NET 8.0 - this whole thing is folded to a just `ret const` including accesses to static arrays by index (via https://github.com/dotnet/runtime/pull/81005) No magic intrinsics involved
This PR is the last missing piece to fold new DateTime(2023, 1, 22) call into just a constant in JIT. While there is not much value in folding DateTime specifically, the PR(s) improves constant fol...
JCStress now knows about NUMA configuration and is able to run the variable-affinity tests knowing that. I.e. it now automatically covers the question "What happens if we run two actor threads on different NUMA nodes?"
Example detected config on TR 3970X:
Like every second week, I blogged again, this week on AsyncGetStackTrace, a better stack trace API for the JVM: https://mostlynerdless.de/blog/2023/01/19/asyncgetstacktrace-a-better-stack-trace-api-for-the-jvm/
I will also talk on this topic at FOSDEM, and I am happy to meet people there.
🐦🔗: https://twitter.com/parttimen3rd/status/1616090222810333184
Wouldn't you like to be able to allocate an infinite amount of memory on your machine?
Read my latest post on how we almost can get there allocating 64 terabytes on a laptop!
Java 20: An Almost Infinite Memory Segment Allocator
http://minborgsjavapot.blogspot.com/2023/01/java-20-almost-infinite-memory-segment.html
See my blog post for explanation of the phenomenon
https://jpbempel.github.io/2020/05/22/startup-containers-tieredcompilation.html
Here's the comparison of my Spring Boot app startup time between standard and native compilation depending on the CPU limit (from 1 core to 0.1 core). I'm testing on Kubernetes.
🐦🔗: https://twitter.com/piotr_minkowski/status/1615456665729474567
Startup, containers & Tiered Compilation Introduction Since the introduction of containers, the way we are running JVM applications has changed a lot. In the same time, microservices architecture rises and the usage of container orchestration like Mesos or Kubernetes has transformed our approach of deployment. With all of that, I end up encountering people configuring a container with less than 1 core. Yes, with docker, for example, we can specify that the container will run with a fraction of a core (called millicores in Kubernetes). The kernel is dealing with that using cpu scheduling quota (see CFS bandwidth). If it could make sense to put 1.5 or 2.5 on an already multi-core machine, putting less than a core has different implications…
Extended OpenJDK JOL (https://github.com/openjdk/jol) to simulate current and future Lilliput project better. You can feed it classes or heapdumps to get an total size estimate under different layout strategies: https://gist.github.com/shipilev/ce53bccd9142ca4f56c703e9df2a8600
@[email protected] @[email protected] My first step for checking inlining would be to visualise in #JITWatch where you can view annotated invoke bytecodes and the tooltips tell you inlined/vcall. Next I would use JFR method profiling for top itable/vtable stubs (get a good sized profile).
🐦🔗: https://twitter.com/chriswhocodes/status/1612933176900329473
“@forked_franz @p_bouda My first step for checking inlining would be to visualise in #JITWatch where you can view annotated invoke bytecodes and the tooltips tell you inlined/vcall. Next I would use JFR method profiling for top itable/vtable stubs (get a good sized profile).”
