Stories by matt_d

HartBreaker: Deterministic Fuzzing of Multi-Hart RISC-V CPUs

Tuning LLVM's SLP Vectorizer Cost Model

A Friendly Tour of Substructural, Uniqueness, Ownership, Capabilities and more!

FlashLib: Bringing Flash Magic to Classical Machine Learning Operators

FML-Bench: A Controlled Study of AI Research Agent Strategies

Finding deadlocks in CuTe kernels with SPIN

A Case for Tracing Based DSL Kernel Languages

You don't need all the LLM benchmarks

Elusive order of async GPU kernels: scheduling, abstractions, DSL implications

MileStone: A Multi-Objective Compiler Phase Ordering Framework

SSV: Sparse Speculative Verification for Efficient LLM Inference

Characterizing Real-World Bugs in Tile Programs for Automated Bug Detection

Characterization of machine learning compilers for LLM inference on NVIDIA GPUs

Chip design from the bottom up – Reiner Pope [video]

LT2: Linear-Time Looped Transformers

Event Tensor: A Unified Abstraction for Compiling Dynamic Megakernel

PopPy: Opportunistically Exploiting Parallelism in Python Compound AI Apps

CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs

[RFC] Open Access to Standards Documents – LLVM Project

Curly braces: An evolution of UNIX and C

2 3 4 5 6 7 8 9 10 11