Signals 4 · free daily AI digest

Dissecting GPU Utilization for LLM Inference on Nvidia Hopper

Paper recorded by Signals 4 on 2026-09-11 in cs.LG. Abstract reproduced from arXiv; link to the original below.

Published 2026-09-11 on arXiv · recorded by Signals 4 on 2026-09-14

Category: cs.LG · 机器学习 · first seen 2026-09-14

Abstract

A single SM utilization percentage can make an LLM inference workload look compute-saturated while hiding how much useful work is being done. The problem is not that the counter is wrong, but that it collapses several different mechanisms into one number. This is most severe during decode, where each request contributes only one new token and dense projection GEMMs become small-row matrix multipli

Read on arXiv →

#66 most recent of 215 cs.LG papers we have recorded · ↑ newer: A Large-Scale AIS Dataset from Finnish Water · ↓ older: Data Scarcity and Model Sparsity: Mixtures-of-Experts Overfit More to
Cite this page: Dissecting GPU Utilization for LLM Inference on Nvidia Hopper: the #66 most recent of 215 cs.LG papers we have recorded (as of 2026-09-11). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/dissecting-gpu-utilization-for-llm-inference-on-nvidia-hopper.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable: papers.json
Related: More cs.LG papers · arXiv signals · All papers · Today in AI
Get 4 AI signals a day by email — free.
Subscribe free → See all plans →
Get 4 AI signals a day by email — free
All models · All repos · By company · Daily editions