Code Reflection in Action - Translating Java to SPIR-V

Presented by Steve Dohrmann - Senior Staff Software Engineer (Intel Corporation) during the JVM Language Summit (August 2024 - Santa Clara, CA).

This video describes the use of the Babylon code reflection API in translating Java code, first to a SPIR-V code reflection dialect, and then to a SPIR-V binary module for execution on a GPU. SPIR-V is an IR specializing in representing parallel compute and graphics workloads. It is used by graphics libraries such as Vulcan, and GPU compute frameworks such as OpenCL and oneAPI.

Make sure to check the    • JVM Language Summit 2024   playlist.

Resources
• Project Babylon ➤ https://github.com/openjdk/babylon/
• SPIR Overview ➤ https://www.khronos.org/spir/
• Java to SPIR-V Example ➤ https://github.com/openjdk/babylon/tr...
• Beehive SPIR-V Toolkit ➤ https://github.com/beehive-lab/beehiv...
• Intel's Level Zero library (Intel GPU low-level access) ➤ https://github.com/oneapi-src/level-zero
• Triton Example ➤ https://github.com/openjdk/babylon/tr...
• Exploring Triton GPU programming for neural networks in Java ➤ https://openjdk.org/projects/babylon/...
• Heterogenous Accelerator Toolkit (HAT) ➤ https://github.com/openjdk/babylon/tr...
• Inside Java ➤ https://inside.java/tag/babylon
• Dev.Java ➤ https://dev.java
• JVMLS ➤ https://openjdk.org/projects/mlvm/jvm...

Tags: #Java #OpenJDK #GPU