Stories by matt_d

Real or Slop? – Programming Languages Papers Edition

Mamba-3

EvoX: Letting AI Evolve Its Own Evolution Process

Native DSLs Ops in PyTorch

Flash-KMeans: Fast and Memory-Efficient Exact K-Means

Gluon: Explicit Performance

Block Number Formats are (Still!) Direction Preservers

cuTile Rust: a safe, tile-based kernel programming DSL for Rust

KernelBlaster: A framework for in context learning for code optimization

Demystifying and Improving Lazy Promotion in Cache Eviction [pdf]

Journeying through Optimization with Heuristics [video]

To Sparsify or to Quantize: A Hardware Architecture View

Efficient sparse computations using linear algebra aware compilers (2025)

A Field Guide to Reward Hacking in AI Kernel Generation

AI and the Mixed-Consistency Future

FIDES: End-to-end Compartments for Mixed-language Systems [pdf]

Practical Type Inference: High‑Throughput Recovery of Real‑World Types

Idempotent Slices with Applications to Code-Size Reduction

Designing AI Chip Hardware and Software

Refinement Modeling and Verification of RISC-V Assembly Using Knuckledragger

1 2 3 4 5 6 7 8 9 10