SOFTWARE ARCHITECT AND INDEPENDENT TECHNICAL TRAINER
HotSpot JVM Performance Tuning Guidelines Content Intro Memory Heap Off-Heap Metaspace CodeCache Direct Buffers ClassLoader Dynamic Class-Data Sharing Just-In-Time Compiler Tiered Mode: C1+C2 Graal JIT Threads Garbage Collectors Serial Garbage Collector Parallel Garbage Collector Concurrent Mark Sweep Garbage Collector G1 Garbage Collector Z Garbage Collector Shenandoah Garbage Collector Epsilon Garbage Collector Container References Intro Tuning…
Introduction In this blog post, David Leitner and I want to give you an insight, based on our experience from the past years of speaking at international conferences (e.g. more than 50 conferences over the last couple of years), about some less obvious but quite significative things you should be aware of when submitting a…
Context The current article describes a series of Java Virtual Machine (JVM) Garbage Collectors (GC) micro-benchmarks and their results, using a different set of patterns. For the current issue, I included all Garbage Collectors from AdoptOpenJDK 64-Bit Server VM version 13 (build 13+33): Serial GC Parallel/ParallelOld GC (starting Java 7u4 Parallel GC and ParallelOld GC…
Key takeaways the methodology, whatever that is (e.g. SCRUM, SAFE, etc) should not impose the necessity of having an architect in the team, but rather the real project needs. in the case of greenfield projects, for projects with a lot of external dependencies, projects under intensive development that involve architectural decisions, technology or framework elicitation,…
Motivation In the current post, I would like to explore different logging patterns and to identify which one is more efficient in terms of performance. Since logging is spread across almost every application, it is important to be aware of what is the most optimal pattern to rely on when writing logging messages. In this…
Context and Motivation The current article describes a series of Java Virtual Machine (JVM) Just In Time (JIT) Compilers micro-benchmarks and their results, relying on different optimization patterns or intrinsics support. For the current issue, I included two compilers from OpenJDK 64-Bit Server VM version 13 (build 13+33) C1/C2 JIT Graal JIT From my point of…
According to the licensing policy, I would like to mention this article was originally published on InfoQ. Key Takeaways While the role and responsibilities of a software architect can be seen as contradictory to the values of the Manifesto for Agile Software Development, a good architect finds techniques that support an agile development team. Working software over…
Context and Motivation The current article describes a series of Java Virtual Machine (JVM) Just In Time (JIT) Compilers benchmarks and their results, relying on different optimization patterns. For the current issue I included only two compilers: Oracle HotSpot C1/C2 JIT Oracle HotSpot Graal JIT In the future, I might extend it to Eclipse OpenJ9, Azul…
Motivation In the current post, I would like to investigate how megamorphic call-sites are optimized in Kotlin as a result of explicit inlining (using the explicit inline modifier) and what is the performance gain out of it. The idea crossed my mind while I was reading the official kotlinlang.org website which mentions “inlining may cause the generated…
Motivation Current post tackles the problem of chaining (or linking) multiple lambda calls which seem to be differently optimized by the HotSpot Just In Time Compiler C2 (i.e. JIT C2) and GraalVM JIT Compiler. In this regard, I would propose to start detailing the problem, to run and check the benchmark results, to find…