Chris Lattner: Compilers, LLVM, Swift, TPU, and ML Accelerators | Lex Fridman Podcast #21
Why Hacker News recommends it
HN commenters point to it as an interesting conversation with the creator of Swift and LLVM, especially for compilers and ML compilation.
What it is
Lex Fridman interviews Chris Lattner, creator of LLVM, Clang and Swift, about compilers, Swift for TensorFlow, TPUs and ML accelerators. The talk covers how hardware and software meet to produce efficient code.
Recommended by 2 HN commenters across 2 different months, 2019 to 2025.
What Hacker News said
Recommends mwattsun · 2021-12 · on “Apple’s use of Swift and SwiftUI in iOS 15”
I always wondered why the creator of Rust switched to Swift and just now found my answer by Graydon Hoare himself: [LINK] tl;dr it has nothing to do with the relative merits of the two languages --- Edit: I watched a couple of Lex Fridman podcasts this week with Chris Lattner (creator of Swift) that I thought were interesting Chris Lattner: Compilers, LLVM, Swift, TPU, and ML Accelerators | Lex Fridman Podcast #21 [this video] Chris Lattner: The Future of Computing and Programming Languages | Lex Fridman Podcast #131 [LINK]
Recommends tony · 2025-09 · on “ML needs a new programming language – Interview with Chris Lattner”
Chris Lattner also has a very good episode on Lex Fridman (Episode 381, June 2nd, 2023): - [LINK] - [LINK] - [LINK] He has two other episodes on the show: - [LINK] (Episode 131, October 18th, 2020) - [this video] (Episode 21, May 13th, 2019)
Cites cs702 · 2019-09 · on “Differentiable Programming Mega-Proposal”
How does this relate to all the work that Christ Lattner et al have been doing at Google with Swift, MLIR, etc.?[a] Is this... a separate, parallel, more encompassing proposal? Is there any coordination between these two groups? -- [a] [this video]
Cites mynegation · 2019-09 · on “Ask HN: Compiler Engineers, what would you advise new grads/students to learn?”
> Aren't there already many ML libraries for compiled languages out there or are you referring to something else? Something else. If your program is a well-defined set of operations (basically matrix operations for ML), you can optimize the whole specific program instead of calling bunch of individually optimized functions from the library and target a specific hardware (e.g. given GPU). Check for example Chris Lattner’s interview [1] and presentations on MLIR, or proceedings of C4ML workshop [2]. Application in distributed systems: from the top of my head - protobufs, compilation of Erlang and Elixir to BEAM, dask. [1] [this video] [2] [LINK]