2Making Equality Saturation Usable for Developing Vectorized Compilers (opens in new tab)(dl.acm.org)1matt_d7h ago0Save
3Reading AI Model Compilation in MLIR Through the Lens of Formal Theories (opens in new tab)(arxiv.org)arXiv2matt_d7h ago0Save
4Liveness Proofs in Veil, Part I: The First Step (opens in new tab)(proofsandintuitions.net)2matt_d13h ago0Save
5ParallelKernelBench: Can LLMs write fast multi-GPU kernels? (opens in new tab)(github.com)GitHub3matt_d18h ago0Save
6LXM: Better Splittable Pseudorandom Number Generators (and Almost as Fast) [video] (opens in new tab)(youtube.com)Video2matt_d1d ago0Save
8PICO: Performance Insights for Collective Operations (opens in new tab)(ieeexplore.ieee.org)4matt_d1d ago0Save
11VoltanaLLM: Energy-Efficient LLM Serving (opens in new tab)(supercomputing-system-ai-lab.github.io)3matt_d2d ago0Save
12Why Software Requirements Get Easier in an AI Economy (opens in new tab)(stng.substack.com)4matt_d2d ago0Save
14Inference Compute Shapes Frontier LLM Evaluation (opens in new tab)(arxiv.org)arXiv2matt_d2d ago0Save
15Concordia: JIT-Compiled Persistent-Kernel Checkpt for Fault-Tolerant Inference (opens in new tab)(arxiv.org)arXiv2matt_d2d ago0Save
