RSVP Please complete this form by EOD Friday, 12/02, if you’d like food.
Monday, Dec 5th
Speaker: Maksim Levental
Room: JCL 298
Many people believe that MLIR stands for Max Levental’s IR. While highly plausible, it is (much to the chagrin of the speaker) not the case. So what is MLIR? Is it a machine learning compiler? Is it Google’s latest way of making TensorFlow a pain to use? Is it Skynet come to fruition? In this informal talk I’ll discuss what MLIR really is, specifically highlighting how it differs from the aforementioned possibilities, and showcase some of the things it can be used for. In particular, I’ll give a tutorial on how one can get started using and developing MLIR. No prior compiler experience necessary. Bring a laptop if you want to play along.
Max Levental is a (n+1)th year PhD student at University of Chicago working on compilery things, for science, with Ian Foster and the Chards (Kyle and Ryan). He hails from the nether regions of Eastern Europe (Moldova) by way of the nether regions of USA (Florida). He knows enough about some things (compilers, FPGAs, deep learning) to be dangerous but primarily uses that knowledge to annoy his group mates (and his wife).