Stories by matt_d

Challenges and Research Directions for Large Language Model Inference Hardware

Library Liberation-Competitive Performance Through Compiler-Composed Nanokernels

Non-Traditional Profiling: "you can just put whatever you want in a jitdump"

Triton Extensions: a framework for developing and building compiler extensions

FlashInfer-Bench: Building the Virtuous Cycle for AI-Driven LLM Systems

High-Performance DBMSs with io_uring: When and How to use it

Are DBMS Researchers Making Correct Assumptions about Transaction Workloads?

vLLM: An Efficient Inference Engine for Large Language Models

Microarchitecture: What Happens Beneath – Matt Godbolt [video]

SMTMSMT: Gluing Together CVC5 and Z3 Nelson Oppen Style

Tilus: A Tile-Level GPGPU Programming Language for Low-Precision Computation

Optimal Software Pipelining and Warp Specialization for Tensor Core GPUs

Oral History of Jeffrey Ullman [video]

CPU Autoscaling with a Kernel of Truth

ACM Transactions on Programming Languages & Systems: New Year, New Paper Tracks

An Empirical Study of Bugs in the rustc Compiler (OOPSLA 2025) [video]

A "Ready-to-Use" Template for LLVM Out-of-Tree Passes

Mini-SGLang: Efficient Inference Engine in a Nutshell

FrontierCS: Evolving Challenges for Evolving Intelligence

svc-hook: hooking system calls on ARM64 by binary rewriting

1
2
3
4
5
6
7
8
9
10