Petr Bouda (@p_bouda) header
Petr Bouda (@p_bouda) profile picture

Petr Bouda

@p_bouda

No longer here: please follow me on https://nitter.gq/t.co/o4b6Kfqjy3

 Prague
dzone.com/users/1131757/BoudaP.html
Joined March 2011
  • Tweets1.3K
  • Following92
  • Followers502
69Media
Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)Photo by Petr Bouda (@p_bouda)
🔥 Part 2 of my Jeffrey series on DZone — differential flame graphs. Set a secondary profile and diff two recordings in one graph: red = regression, green = improvement, deep shades = frames that only exist on one side. No more eyeballing two graphs. dzone.com/articles/java-diff… #Java
Differential Flamegraphs in Java in Jeffrey Microscope

Differential Flamegraphs in Java in Jeffrey Microscope

How to set up a secondary profile and pinpoint the precise frames responsible for a performance change between two JFR recordings.

dzone.com
1
2
109
🔥 New on DZone — part 1 of my Jeffrey series, an open-source performance analyzer for Java. Spin up Microscope, drop in a recording, read the flame graph: CPU, allocation, wall-clock. Plus "Copy for AI" to hand hot paths to an LLM. dzone.com/articles/jeffrey-j… #Java
Jeffrey Microscope for Generating Flame Graphs in Java

Jeffrey Microscope for Generating Flame Graphs in Java

Jeffrey Microscope is a deep analyzer for JFR recordings, with a particular focus on flamegraphs generated from stacktrace-based events

dzone.com
1
2
413
🔥 Jeffrey 0.5 is HERE! 🔥 ✨ Big Upgrade of Java performance detective. Ready to level up your Java performance game?✨ Look at the current features: jeffrey-analyst.cafe/#featur… See it in action: jeffrey-analyst.cafe/#/tour-… #Java #Performance #JFR #OpenJDK
3
113
No longer here: please, follow me on bsky.app/profile/pbouda.bsky…
1
76
🚀 Jeffrey 0.4 - Native Memory Leaks 🚀 jeffrey-analyst.cafe/posts/0…
2
5
280
🚀 Jeffrey 0.3 Anouncement 🚀 Try it out and give feedback. Plenty of new features, some of them are experimental that wait for your real workloads! Guardians / Wall-Clock / Thread View / and others jeffrey-analyst.cafe/posts/0…
3
1
4
765
Petr Bouda retweeted
Wasm3 (@wasm3_engine) profile picture
Wasm3@wasm3_engine18 Dec 2023
I regret to inform the community that since my own house was destroyed by russians who invaded my country, Wasm3 has entered a minimal maintenance phase.
At this time, I am unable to continue the development of new features. However, I am committed to keeping the project alive and will actively review and merge incoming Pull Requests. I deeply appreciate your understanding and support during this difficult period. Your contributions to Wasm3 are now more valuable than ever.

ALT At this time, I am unable to continue the development of new features. However, I am committed to keeping the project alive and will actively review and merge incoming Pull Requests. I deeply appreciate your understanding and support during this difficult period. Your contributions to Wasm3 are now more valuable than ever.

58
864
50
4,079
660,703
1st edition was fantastic, 2nd will surely be awesome as well 🙂
Image from Petr Bouda (@p_bouda)
4
106
📢 It's here! The second edition of my book "Performance Analysis and Tuning on Modern CPUs" is out NOW! Dive into low-level optimizations and learn to write fast code like a pro. Grab your copy: amzn.to/4fsroBs Please RT to spread the word! #NewBook #BookRelease 🧵1/3
Performance Analysis and Tuning on Modern CPUs: Learn to write fast software like a pro

Performance Analysis and Tuning on Modern CPUs: Learn to write fast software like a pro

Modern software is massively inefficient. Modern CPUs typically run poorly optimized code, which consumes more power than it could have consumed and contributes to carbon emissions and other enviro...

amazon.com
15
111
9
441
82,768
Replying to @p_bouda @richardstartin
github.com/async-profiler/as… This is what async-profiler does (with -e cpu) and there, what's the kernel does with that SW event (which means -> skidding without PEBs help)
why async-profiler cpu event less than perf · Issue #801 · async-profiler/async-profiler

why async-profiler cpu event less than perf · Issue #801 · async-profiler/async-profiler

i started perf and async-profiler on same process. the higher line is perf and lower line is async profiler: and sees async-profiler always less than perf. the perf result: the async profiler result:

github.com
1
1
152
JEP 450 is not the end of Project Lilliput, btw. We are already busy working on 4-byte headers (down from 12 bytes). In-fact we already have a fully-working prototype. ☺️
Roman Kennke 🗽 (@rkennke) profile pictureRoman Kennke 🗽@rkennke8 Nov 2024
And ... it's landed! 🥳🥳 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! 🙏🙌
8
18
89
7,317
🥳 JEP proposed to target JDK 24: 450: Compact Object Headers (Experimental) 🎉 mail.openjdk.org/pipermail/j…
3
24
1
91
9,557
Many JFR events don't have any descriptions, so sapmachine.io/jfrevents/ now includes code snippets from the events usage in OpenJDK:
Image from Johannes Bechberger (@parttimen3rd)
2
12
27
2,169
JEP 8337789 ("CPU Time Profiling for JFR") updated: Status changed from "Draft" to "Submitted". See openjdk.org/jeps/8337789
1
4
1
10
1,458
...and it is in! Already in bleeding edge builds: builds.shipilev.net/openjdk-…
3
7
1,383
Replying to @BachorikJ @nitsanw
I've spent some time streamlining the generated code and reducing the chained calls to MethodHandles - github.com/jbachorik/jafar/p…. Now, it can do the 'benchmark' on 1.79GiB in ~2s walltime (~14s cpu time) beating all other parsers by healthy margin. #java #jfr #jafar
Rework the handler code generation by jbachorik · Pull Request #7 · btraceio/jafar

Rework the handler code generation by jbachorik · Pull Request #7 · btraceio/jafar

Do not rely on generic deserialization methods accessed via method handles for reading, and most importantly skipping values. Generate in-line bytecode to fully deserialize or skip full type instan...

github.com
2
7
432
My JFR Events Collection got an update: It now includes all events from JDK 23 and 24 (thank you to @p_bouda for the bug report) and uses a new version of the Renaissance benchmark suite for the example files. You can find it at sapmachine.io/jfrevents/
Image from Johannes Bechberger (@parttimen3rd)
2
4
13
888
Hi @parttimen3rd 🙂 I'm using your fantastic toolhttps://sap.github.io/SapMachine/jfrevents, however, it looks like version 23 miss some events in comparison to 21. e.g. section `Java Application` 🙂
1
1
1
2
1,297
Younger brother of mine is growing to a powerful beast! Let me know if you have any interesting pattern consuming your resources in JFR recordings.
Image from Jeffrey, The JFR Analyst (@jeffrey_analyst)
Image from Jeffrey, The JFR Analyst (@jeffrey_analyst)
2
2
213
The trilogy of #java FFM API performance improvements is now complete in Java 24! A while ago, we integrated github.com/openjdk/jdk/pull/… which improves the performance of MemorySegment::fill This improves performance on all supported Java platforms, in particular on Aarch64.
Image from Per-Åke Minborg (@PMinborg)
1
4
23
1,607
Load more