[In-Person - March] Optimizing Java for Modern Processors: Vector API

## Details

Location:
Oracle Tech Hub
Training room 2276, Block A
Marathahalli - Sarjapur Outer Ring Rd,
Kadubeesanahalli, Bengaluru -560103

-- IF YOU DECIDED NOT TO JOIN, PLEASE RSVP TO NO --
-- ITS A FREE EVENT, RSVP before 13th Mar 2025 --
-- GOV ID CARD IS MANDATORY TO ENTER --

Talk #1: Optimizing Java for Modern Processors: Harnessing SIMD with the Vector API
Time: 10.30am to 11.30 am

The Java Vector API, introduced as an incubator feature in JDK 16, allows developers to leverage SIMD (Single Instruction, Multiple Data) capabilities in modern processors while maintaining Java’s platform independence through a hardware-agnostic abstraction. By utilizing hardware acceleration, it enables explicit vectorized operations across supported architectures, significantly boosting performance compared to scalar computations. The API simplifies low-level vectorization by supporting automatic loop unrolling and data parallelism, leading to improved throughput in compute-intensive applications such as database analytical engines, vector searches, cryptography, and machine learning. This talk provides insights into these new APIs and, as a testament to their performance boost, will share the benefits observed in data analytics filtering operations with a database analytical engine "Presto" (Not a demo - but highlight the results observed).

Speaker(s) :
Abhijit Kulkarni

Talk #2: Scaling Java for AI Workloads: what and why of JVM innovations.
Time: 11.30 am to 12.30 pm

Java has long powered enterprise applications, but AI workloads demand specialized optimizations such as efficient memory structures, seamless hardware acceleration, and adaptive execution models. How JVM is evolving to bridge this gap? this session provides a 25000 feet view of the cutting-edge JVM innovations enabling Java for AI-driven workloads. I plan to touch upon value types and specialized generics to reduce memory fragmentation and efficient tensor operations, ffi & memory api for zero-copy interop with GPU/TPU-accelerated AI libraries such as tensorflow optimized vectorization & SIMD acceleration for high-speed AI computations, enhancing AI data processing and transformations, scaling concurrent AI workloads efficiently. Developers will learn practical strategies to optimize Java for AI, and appreciate how JVM advancements align with next-gen AI infrastructure.

Speaker(s) :
Varada M

Talk #3: JEP 492: Flexible Constructor Bodies
Time: 12:30 pm to 1.00 pm
JEP 492 introduces flexible constructor bodies, enabling statements to appear before an explicit constructor invocation (i.e., super(..) or this(..)). This enhancement addresses the limitations of the current constructor structure, where statements must follow the constructor call. By allowing statements before the invocation, JEP 492 improves code flexibility and readability, reducing the need for workarounds.

Speaker(s) : Ravi Gupta

Break#: Snacks and Networking
Time: 1:15pm onwards