• Java can be faster than Rust

    From ram@ram@zedat.fu-berlin.de (Stefan Ram) to comp.lang.java.programmer on Mon Oct 19 20:16:50 2026
    From Newsgroup: comp.lang.java.programmer

    | The tech community is actively discussing the Campfire framework - a
    | project originally written in Rust that saw significant performance
    | gains after being ported to Java. While this outcome surprised many
    | given Rust's bare-metal execution model, it highlights Java's robust
    | runtime optimizations and sets the stage for the highly anticipated
    | JEP 544 (Ahead-of-Time Code Compilation).
    |
    | The Campfire Context: Why Java Can Beat Rust
    |
    | The performance advantage in the Java port of Campfire stems from
    | specific architectural realities rather than raw CPU processing speed:
    |
    | o Garbage Collection vs. Explicit Memory Management: Rust relies on
    | strict ownership and explicit allocation/deallocation. In highly
    | dynamic, short-lived object scenarios, Java's modern garbage
    | collectors (like ZGC or G1) can allocate memory incredibly fast via
    | thread-local allocation buffers (TLABs), deferring the cleanup cost.
    |
    | o Just-In-Time (JIT) Dynamic Optimization: Java's C2 compiler profiles
    | the application while it runs. It performs deoptimization, aggres-
    | sive inlining, and escape analysis based on real-world runtime be-
    | havior - optimizations that a static, ahead-of-time compiler like
    | Rust's rustc cannot always predict.
    |
    | JEP 544: Ahead-of-Time (AOT) Code Compilation
    |
    | JEP 544 introduces a formalized framework for Ahead-of-Time (AOT)
    | compilation into the standard OpenJDK pipeline. It builds heavily on
    | the groundwork laid by Project Leyden to address Java's historic
    | weaknesses without sacrificing its runtime strengths.
    |
    | To understand how JEP 544 shifts the paradigm, it is best evaluated
    | across key multi-attribute technical metrics:
    |
    | .--------------------------------------------------------------------.
    | | JEP 544 Architectural Comparison |
    | |--------------------------------------------------------------------|
    | |Architectural | Traditional JIT | JEP 544 (AOT | Impact on |
    | |Metric | (HotSpot C2) | Compilation) | Frameworks like |
    | | | | | Campfire |
    | |--------------------------------------------------------------------|
    | |Startup Time | Slow. The JVM | Instantaneous. | Crucial for |
    | |& "Warmup" | must interpret | Pre-compiled | cloud-native |
    | | | bytecode and | native machine | scaling and |
    | | | compile hot | code executes | microservices. |
    | | | methods | immediately | |
    | | | iteratively. | upon boot. | |
    | |--------------------------------------------------------------------|
    | |Memory | High. Requires | Low. Strips out | Lowers |
    | |Footprint | extra memory to | heavy runtime | infrastructure |
    | |(RSS) | store the JIT | compilation | costs |
    | | | compiler | machinery, | significantly. |
    | | | infrastructure. | reducing base | |
    | | | | memory usage. | |
    | |--------------------------------------------------------------------|
    | |Peak | Extremely High. | High to Very | Bridges the gap |
    | |Throughput | Optimized | High. Uses | between static |
    | | | continuously | stored | binaries and |
    | | | using live | profiling data | dynamic |
    | | | production | (PGO) to match | optimization. |
    | | | telemetry. | JIT performance | |
    | | | | closely. | |
    | '--------------------------------------------------------------------'
    |
    | Key Thoughts on JEP 544's Strategic Approach
    |
    | 1. It Avoids the "Closed-World" Trap
    |
    | Unlike GraalVM Native Image, which enforces a strict "closed-world
    | assumption" (restricting dynamic class loading and reflection), JEP
    | 544 aims for a flexible AOT model. It allows developers to compile
    | critical parts of the application ahead of time while maintaining
    | standard Java compatibility.
    |
    | 2. Profile-Guided Optimization (PGO) is Mandatory
    |
    | Static AOT code is inherently less optimized than dynamic JIT code
    | because it lacks runtime context. JEP 544 solves this by leaning into
    | Profile-Guided Optimization. By feeding training run data into the AOT
    | compiler, the resulting binary retains the aggressive optimizations
    | (like speculative inlining) that allowed the Campfire Java port to
    | outpace Rust.
    |
    | 3. The Future is Hybrid
    |
    | The ultimate goal of JEP 544 is not necessarily to replace JIT
    | entirely, but to create a hybrid deployment model. Applications can
    | boot instantly using AOT-compiled code, and as the application warms
    | up, the JIT compiler can selectively re-compile performance-critical
    | hot spots to achieve peak throughput.

    Lines marked with "| " come from my editing, where I start by writing
    prompts for the chatbot and then edit the generated texts and format
    them for Usenet. The chatbot might make mistakes and misrepresent
    or invent facts.


    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Mild Shock@janburse@fastmail.fm to comp.lang.java.programmer on Mon Oct 19 23:25:31 2026
    From Newsgroup: comp.lang.java.programmer

    Hi,

    Yeah interesting. Java the old lady of 20-30 years,
    having features like maxMemory() that others lack?

    Bye

    BTW: In some corners people don't do PGO anymore
    because GCC23 has become better or something.

    Stefan Ram schrieb:
    | The tech community is actively discussing the Campfire framework - a
    | project originally written in Rust that saw significant performance
    | gains after being ported to Java. While this outcome surprised many
    | given Rust's bare-metal execution model, it highlights Java's robust
    | runtime optimizations and sets the stage for the highly anticipated
    | JEP 544 (Ahead-of-Time Code Compilation).
    |
    | The Campfire Context: Why Java Can Beat Rust
    |
    | The performance advantage in the Java port of Campfire stems from
    | specific architectural realities rather than raw CPU processing speed:
    |
    | o Garbage Collection vs. Explicit Memory Management: Rust relies on
    | strict ownership and explicit allocation/deallocation. In highly
    | dynamic, short-lived object scenarios, Java's modern garbage
    | collectors (like ZGC or G1) can allocate memory incredibly fast via
    | thread-local allocation buffers (TLABs), deferring the cleanup cost.
    |
    | o Just-In-Time (JIT) Dynamic Optimization: Java's C2 compiler profiles
    | the application while it runs. It performs deoptimization, aggres-
    | sive inlining, and escape analysis based on real-world runtime be-
    | havior - optimizations that a static, ahead-of-time compiler like
    | Rust's rustc cannot always predict.
    |
    | JEP 544: Ahead-of-Time (AOT) Code Compilation
    |
    | JEP 544 introduces a formalized framework for Ahead-of-Time (AOT)
    | compilation into the standard OpenJDK pipeline. It builds heavily on
    | the groundwork laid by Project Leyden to address Java's historic
    | weaknesses without sacrificing its runtime strengths.
    |
    | To understand how JEP 544 shifts the paradigm, it is best evaluated
    | across key multi-attribute technical metrics:
    |
    | .--------------------------------------------------------------------.
    | | JEP 544 Architectural Comparison |
    | |--------------------------------------------------------------------|
    | |Architectural | Traditional JIT | JEP 544 (AOT | Impact on |
    | |Metric | (HotSpot C2) | Compilation) | Frameworks like |
    | | | | | Campfire |
    | |--------------------------------------------------------------------|
    | |Startup Time | Slow. The JVM | Instantaneous. | Crucial for |
    | |& "Warmup" | must interpret | Pre-compiled | cloud-native |
    | | | bytecode and | native machine | scaling and |
    | | | compile hot | code executes | microservices. |
    | | | methods | immediately | |
    | | | iteratively. | upon boot. | |
    | |--------------------------------------------------------------------|
    | |Memory | High. Requires | Low. Strips out | Lowers |
    | |Footprint | extra memory to | heavy runtime | infrastructure |
    | |(RSS) | store the JIT | compilation | costs |
    | | | compiler | machinery, | significantly. |
    | | | infrastructure. | reducing base | |
    | | | | memory usage. | |
    | |--------------------------------------------------------------------|
    | |Peak | Extremely High. | High to Very | Bridges the gap |
    | |Throughput | Optimized | High. Uses | between static |
    | | | continuously | stored | binaries and |
    | | | using live | profiling data | dynamic |
    | | | production | (PGO) to match | optimization. |
    | | | telemetry. | JIT performance | |
    | | | | closely. | |
    | '--------------------------------------------------------------------'
    |
    | Key Thoughts on JEP 544's Strategic Approach
    |
    | 1. It Avoids the "Closed-World" Trap
    |
    | Unlike GraalVM Native Image, which enforces a strict "closed-world
    | assumption" (restricting dynamic class loading and reflection), JEP
    | 544 aims for a flexible AOT model. It allows developers to compile
    | critical parts of the application ahead of time while maintaining
    | standard Java compatibility.
    |
    | 2. Profile-Guided Optimization (PGO) is Mandatory
    |
    | Static AOT code is inherently less optimized than dynamic JIT code
    | because it lacks runtime context. JEP 544 solves this by leaning into
    | Profile-Guided Optimization. By feeding training run data into the AOT
    | compiler, the resulting binary retains the aggressive optimizations
    | (like speculative inlining) that allowed the Campfire Java port to
    | outpace Rust.
    |
    | 3. The Future is Hybrid
    |
    | The ultimate goal of JEP 544 is not necessarily to replace JIT
    | entirely, but to create a hybrid deployment model. Applications can
    | boot instantly using AOT-compiled code, and as the application warms
    | up, the JIT compiler can selectively re-compile performance-critical
    | hot spots to achieve peak throughput.

    Lines marked with "| " come from my editing, where I start by writing
    prompts for the chatbot and then edit the generated texts and format
    them for Usenet. The chatbot might make mistakes and misrepresent
    or invent facts.



    --- Synchronet 3.22a-Linux NewsLink 1.2