Using GPUs to Achieve Massive Parallelism in Java 8
Grand Ballroom 220B
|Wednesday from 12:20 PM til 1:10 PM|
Graphic processing units (GPUs) are not limited to traditional scene rendering tasks. They can play a huge role in accelerating applications that have a large number of parallelizable tasks.
Learn how Java can exploit the power of GPUs to optimize high-performance enterprise and technical computing applications such as big data and analytics workloads, through both explicit GPU programming and letting the Java JIT compiler transparently off-load work to the GPU.
This presentation covers the principles and considerations for GPU programming from Java and looks at the software stack and developer tools available. After this talk you will be ready to extract the full power of GPUs from your own application.
We will present a demo showing GPU acceleration and discuss what is coming in the future.
Adam spends his time deep in IBM Java/Apache Spark's codebase and regularly helps customers to use Spark for the first time. He works on Spark quality assurance, optimising Spark performance, fixing bugs in Spark or related projects such as Zeppelin/Hadoop, and has a keen interest in using hardware accelerators to maximise throughput.
He's responsible for delivering IBM's Development Package for Apache Spark and wants to help people get started with Spark, share working code, and learn from other's experiences. Adam also teaches machine learning techniques to IBMers and the basics to children as young as ten, and has an additional interest in analysing data for ourselves gathered from IoT devices.