AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Threshold-Based Exclusive Batching for LLM Inference

Weifang Zhang, Yuzhou Nie, Bowen Pang, Guangrui Ma, Shining Wu

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Distributed LLM inference systems · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use optimized exclusive batching instead of mixed batching on bandwidth-constrained GPUvLLM v1 / Qwen3-8B · WildChat on L40SRPS: 10.36 -> 14.70Improved
use exclusive batching instead of mixed batching on high-bandwidth GPUvLLM v1 / Qwen3-8B · WildChat on B300RPS: 50.47 -> 52.34No clear effect
use EB(k*) instead of v1 on RTX PRO 6000Qwen3-8B · real-world workloadsaverage RPS gain: +7.9%Improved
use EB(k*) instead of v1 on RTX PRO 6000Qwen3-30B-A3B · real-world workloadsaverage RPS gain: +1.4%Mixed
use EB(k*) instead of v1 on H200Qwen3-30B-A3B · real-world workloadsaverage RPS gain: -2.9%Hurt
scale model from Qwen3-8B to Qwen3-30B-A3BEB gain over v1 · RTX PRO 6000average RPS gain: 7.9% -> 1.4%Hurt
use closed-form EB(k*) instead of best fixed-k sweep (ablation)exclusive batching · H200 decode-heavy, N=1024throughput: +8.0%Improved
use closed-form EB(k*) instead of best fixed-k sweep (ablation)exclusive batching · H200 balanced, N=1024throughput: +3.6%Improved
use closed-form EB(k*) instead of best fixed-k sweep (ablation)exclusive batching · H200 prefill-heavy, N=1024throughput: +0.6%No clear effect
add online hybrid EB+ mode switching over v1scheduler · RTX PRO 6000 distribution shiftthroughput: +36.4%Improved
add online hybrid EB+ mode switching over v1scheduler · H200 distribution shiftthroughput: +3.9%Improved
add online hybrid EB+ mode switching over v1scheduler · RTX PRO 6000 concurrency shiftthroughput: +22.6%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy