home.social

#jmh — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #jmh, aggregated by home.social.

  1. 🚀 Call for Contributors – #JVM #Performance #Benchmarks

    If you're interested in contributing to the #JVM #Performance #Benchmarks project - an initiative that gained significant traction in the #Java community through our recent #JDK17 and #JDK21 analyses - check out the repo:

    🔗 github.com/ionutbalosin/jvm-pe

    🧵 DM me or open a PR to get started

    #Java #JVM #OpenJDK #GraalVM #JMH #Performance #OpenSource

  2. Simple O(n^2) vs mildly optimised solution in #Java for #AdventOfCode day 1 part 2 using #jmh.

    Day 2 part 2 was surprisingly tricky to find the edge cases and required unit tests to track them all down.

    Code: github.com/chriswhocodes/aoc20
    Benchmark: github.com/chriswhocodes/aoc20

  3. Thanks for attending my deep dive session, and thanks for all feedback and questions. #Devoxx #JMH

    Slides t.ly/POQnP and if you want to get really crazy with running some examples, you can clone the repository from github.com/Xceptance/jmh-train

    And here is the full recording: youtu.be/u5XC2nriqpQ?si=SB_05N

    Feedback appreciated!

  4. Hope to see you tomorrow!

    For all #1brc fans, we will lay the basic foundation for that exercise, without elaborating on tuning, by seeing the magic of the JVM and hardware... because it will hit us when benchmarking too naively.
    @Devoxx #jmh #1brc

  5. How many joules does it take to execute a single Java method?

    In my latest Growing Green Software blog post, I explore this question and show how we can use JoularJX to break down the energy consumption of a Java program to the method level.

    medium.com/growing-green-softw

    #GreenSoftware #EnergyMeasurement #Java #JMH #JoularJX

  6. "My code is faster than yours... let me prove it to you!" Challenge accepted, François Martin! 🚨

    In this #BaselOne24 talk, you will learn about benchmarking with #JMH. JMH is a #Java harness for building, running, and analyzing nano/micro/milli/macro benchmarks written in Java and other languages targeting the hashtag#JVM.

    Get your tickets at lnkd.in/ggjmzerN.

    #communityrocks #softwaredevelopment #softwareengineering #BaselOne

  7. Как замерять и повышать производительность Java-кода: личный пример с JMH

    Привет, Хабр! Большинство команд разработки так или иначе вовлечены в гонку за производительностью. Если понимать этот показатель как скорость работы системы, то вся деятельность по прокачке производительности — это, по сути, анализ метрик и поиск инструментов, которые эту скорость могут повысить. Меня зовут Александр Певненко , я работаю в СберТехе, компании, которая разрабатывает ПО. Для большинства наших продуктов производительность — критичный фактор, поэтому анализ метрик и оптимизация кода — важная часть цикла разработки. Хочу поделиться личным опытом и мнением обо всём, что касается оптимизации и повышения скорости работы кода, а также нагрузочного тестирования как части этой работы. Тут представлены моя собственная практика использования набора библиотек JMH для нагрузочного тестирования и замеров производительности Java-кода. Всем, кому интересно, добро пожаловать под кат!

    habr.com/ru/companies/sberbank

    #тестирование #jmh

  8. @matari Most of your recorded time will actually have been spent on the System.out.println command. And self-made benchmark timers are notorious for giving bad data, so I recommend JMH instead.

    You're right that String.format is slower than using the + operator, but a JMH benchmark comparison would show the difference to be much greater. (It would also show that String.format allocates a bunch more bytes to the heap.)

    In summary: don't use String.format for simple concatenation.

  9. Today I made the first release of Energy Impact . See github.com/aixigo/sonarqube-ja

    Based on scientific research the usage of and is discouraged due to higher leading to higher |s. 5% |s can be achieved.

    Now I'm searching for ways to measure energy consumption in Microbenchmark Harness () to start working on the higher hanging fruits/recommendations of the paper: doi.org/10.1007/s10664-021-099

  10. The curious C2 case of different training data and sticking to it forever: github.com/rschwietzke/jmh-C2-

    I would appreciate a review, any feedback, and possibly an idea, if this is a bug, a feature, or just one of the "bad luck" cases.

    Appreciated! #java #jvm #jit #jmh

  11. Here are the slides of my first #DevoxxPL talk. I had to rush a little because I wanted to tell you so much in so little time. Thanks for attending. I hope you enjoyed it and you can take something home.

    See you tomorrow for "Parallel Programming is Hard". 1430/230pm in Room 2!

    training.xceptance.com/java/50

    #java #jvm #jmh

  12. Join me today at 1430/230pm in Room 4B for "Microbenchmarking, the art of realizing one is wrong" #DevoxxPL

    I promise code, funny numbers, some #JVM insides and insights, and car analogies.

    #Java #JMH

  13. How to write JMH microbenchmarks for asynchronous functions (tips and tricks). mail.openjdk.org/pipermail/loo
    By Sergey Kuksenko #Java #jmh

  14. JMH 1.36 (github.com/openjdk/jmh) is released (JDK 8 is now minimum, GC profiler bugfix, perfasm profiler improvements): mail.openjdk.org/pipermail/jmh
    #noxp #Java #jmh

  15. Significantly cleaner #JMH results with this improvement to compiler blackholes, allowing standard load optimisations to work: mail.openjdk.org/pipermail/hot #HotSpot #performance #benchmark