From Install to Insight: A Hands-On GPU Lab for Java Developers

Getting a local LLM running on your GPU from inside Java usually stalls at the installation step — which SDK, which backend, and how do you even tell if the GPU is being used? This hands-on lab removes the guesswork.

Working at your own laptop, you'll install and wire up the GPULlama3.java + TornadoVM stack from scratch: adding TornadoVM via Maven Central, and configuring DevoxxGenie inside IntelliJ IDEA as a local, GPU-backed AI assistant.

Once the stack is running, we'll go hands-on with TornadoVM's tooling ecosystem: the TornadoInsight IDE plugin for real-time compatibility checks and in-IDE task testing, TornadoPulse for reading profiling logs, and TornadoViz for visualizing how your code actually maps onto the GPU. Along the way you'll run and modify small exercises to see JIT-compiled Java execute on your hardware, and — if you're on an NVIDIA GPU — capture a basic profile with Nsight Systems.

We'll close by looking ahead at some early, ongoing exploration into what GPU-accelerated local inference could look like baked directly into an IDE assistant (e.g., DevoxxGenie) - more a glimpse of a direction than a finished feature.

No prior GPU programming or TornadoVM experience required. By the end, you'll have a working local GPU-accelerated LLM setup on your own machine, plus the tools to inspect and debug it - a foundation you can carry straight into your own projects.