Cyfra: Scala on GPU
Learn how to accelerate Scala code by orders of magnitude with Cyfra.
There are many libraries that give you access to low-level GPU APIs, but a high-level, functional one is a different story. That’s why we created the Cyfra GPGPU framework. In this talk, we’ll demonstrate how to use it to build high-performance GPU applications with Scala.
In this talk, we will show how to build high-performance Scala GPU applications with Cyfra. First, we will introduce the audience to the current landscape of GPU programming and explain our philosophy. Then, we will dive into our model of type-safe data pipelines and memory management. To complete the picture, we will also showcase a set of real-world applications, including an end-to-end LLM inference engine, written entirely in Scala.
Szymon is an engineer that is working on System-on-chip interconnects at Arteris. At the same time, he contributes to open source projects, such as the Scala compiler and Cyfra. In the past, he was a member of the Scala compiler team, and he was the organiser of Kraków Scala User Group.
Marcin is an engineer specializing in performance, building next generation storage systems. In his spare time, he contributes to Scala open source projects, including Akka Serialization Helper and Cyfra.