The Use of MLIR in Magma

Speaker: Raj Setaluri, PhD Student, Stanford University
Date: April 6, 2022

Our hardware generator language, Magma, has seen increased usage, both in terms of number of users and complexity of designs. With this increased usage, we have seen a corresponding increase on demands of runtime performance, code generation quality, and debuggability. To address these issues, we have implemented a new MLIR-based backend. In this talk we will give an overview of the MLIR compiler toolchain, and discuss how we have leveraged this new backend to improve Magma.