Witch of Git - web/blog/blob - posts/2024/llvm-dev-meeting.md

   1 ---
   2 title: "LLVM Dev Meeting thoughts"
   3 author: Cassie Jones
   4 date: "2024-10-27 11:13:00 America/Los_Angeles"
   5 summary: "I went to the LLVM Dev Meeting this year! My first time at a conference in years, and the first time at one professionally. Here's what it was like:"
   6 ---
   7
   8 This year I went to the LLVM Dev Meeting (Oct 22–24), officially sponsored to do so for work!
   9
  10 It was just across town, but I had to wake up early and it was far enough away that I didn't want to take an hour long bus ride at 7am, so I felt forced to use Lyft.
  11
  12 I was only one of like 4 people total I saw with a mask, which is annoying.
  13 It wasn't very connected with outside, although there was outside seating for all the meals which made things managable for me.
  14 I definitely ended up somewhat dehydrated because of this though.
  15
  16 It was nice to talk to people!
  17 I liked talking with coworkers in a different environment, got to hear new things that way. (I miss the
  18 I liked
  19
  20 Getting to meet [John Regehr](https://john.regehr.org/) in person was very nice!
  21 A highlight of the event because we've talked a bit on twitter and mastodon for ages.
  22
  23 I discovered that my old personal laptop (the 2-core 8GB 2015 MBP I'm writing this post on) takes roughly two full charges to do a clean compile of clang.
  24 I wanted to experiment with something but had a year-old build of LLVM, I started compiling at full battery, discharged to 10%, charged back to full, and discharged back to 20% by the time compilation finally finished.
  25
  26 I got 3 LLVM t-shirts, one is the conference shirt, two were just LLVM merch from the stand.
  27 My favorite is the shirt of Jessica's LLVM dragon book design.
  28
  29 https://mastodon.social/@barrelshifter/108201300354237574
  30
  31 My favorite talk was the one on [Lean-MLIR][], which showed a nice setup for formally verifying peephole optimizations.
  32 They seem to have a [nice online playground setup][lean-mlir-playground], I really want to play around with proving some optimizations with it.
  33
  34 [Lean-MLIR]: https://github.com/opencompl/lean-mlir
  35 [lean-mlir-playground]: https://lean-mlir.grosser.es/