LLVM/Clang Social Darmstadt

Come and join us to share your experiences and interest in LLVM/Clang and related tools.

Agenda
18.45 Doors Open
19.00 Welcome
19.05 "Compiling Machine Learning Models with PyTorch 2.0 and Triton", Lukas Sommer (Codeplay)
20.00 Social w/ food and drinks
22.30 Tentative End
The meetup is hosted in room 7 of building S1|03.

Take the main entrance from Hochschulstraße into the building, then turn left and head down the hallway. The door will be open.

Abstract
Since the first release of ChatGPT to the public in late 2022, AI-powered chat-bots and the underlying technology, large language models (LLM), have created an unprecedented level of interest in artificial intelligence.
A crucial building block to further grow the size of LLMs and improve their capabilities are efficient training and inference on GPUs. In order to leverage the full computational power of these heterogeneous devices, PyTorch introduced an entire compiler stack into the framework, starting with version 2.0.
In this talk, Lukas is going to explain the PyTorch 2.0 compilation stack and what role the Triton language plays in the compilation flow. Lukas is also going to give a quick introduction to MLIR and demonstrate how MLIR is used in the Triton language compiler.