# PowerSlider: Exploiting Phase Asymmetry for LLM Serving under Demand Response

arXiv: https://arxiv.org/abs/2608.21719 · HTML: https://mlsanity.com/paper/2608.21719/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Serving systems and request scheduling, one of 12 directions in Efficient LLM inference and serving.
- Builds on 17 of its 84 references that are in the map.
- First posted on arXiv 22 Aug 2026.

## Placed in

- Efficient LLM inference and serving / [Serving systems and request scheduling](https://mlsanity.com/inference/serving-systems-and-request-scheduling/)

## Builds on (most cited first, 17 of 17)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Convex optimization](https://mlsanity.com/paper/2106.01946/) (2021)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing](https://mlsanity.com/paper/2406.08464/) (2024)
- [Zeus: Understanding and Optimizing GPU Energy Consumption of DNN Training](https://mlsanity.com/paper/2208.06102/) (2022)
- [Carbon Explorer: A Holistic Approach for Designing Carbon Aware Datacenters](https://mlsanity.com/paper/2201.10036/) (2022)
- [TAPAS: Thermal- and Power-Aware Scheduling for LLM Inference in Cloud Platforms](https://mlsanity.com/paper/2501.02600/) (2025)
- [SLO-aware GPU Frequency Scaling for Energy Efficient LLM Inference Serving](https://mlsanity.com/paper/2408.05235/) (2024)
- [EcoServe: Designing Carbon-Aware AI Inference Systems](https://mlsanity.com/paper/2502.05043/) (2025)
- [SLOs-Serve: Optimized Serving of Multi-SLO LLMs](https://mlsanity.com/paper/2504.08784/) (2025)
- [CarbonScaler: Leveraging Cloud Workload Elasticity for Optimizing Carbon-Efficiency](https://mlsanity.com/paper/2302.08681/) (2023)
- [Let's Wait Awhile: How Temporal Workload Shifting Can Reduce Carbon Emissions in the Cloud](https://mlsanity.com/paper/2110.13234/) (2021)
- [Ecovisor: A Virtual Energy System for Carbon-Efficient Applications](https://mlsanity.com/paper/2210.04951/) (2022)
- [The War of the Efficiencies: Understanding the Tension between Carbon and Energy Optimization](https://mlsanity.com/paper/2306.16948/) (2023)

## Built on by (most cited first, 2 of 2)

- [Toward Sustainable Distributed LLM Inference: A Systems Synthesis and Research Agenda for an Energy-, Carbon-, and Cache-Aware llm-d Control Plane](https://mlsanity.com/paper/2609.05565/) (2026)
- [Analytical Power-Aware Provisioning for Prefill-Decode Disaggregated AI Inference](https://mlsanity.com/paper/2609.24639/) (2026)

## Abstract (arXiv)

AI inference clusters are increasingly constrained by instantaneous power, not just energy: grid operators condition new capacity on demand response, imposing time-varying power caps. Existing LLM serving systems optimize a static energy objective or shed fixed priority tiers under load; either way, goodput collapses when the power envelope moves. An LLM pipeline is not a uniform load: compute-bound prefill loses throughput almost linearly with GPU frequency, memory-bound answer decode sustains it down to 0.57× nominal, and reasoning's thinking phase couples KV-cache capacity to scheduling -- so a cap should be steered to where each watt costs the least performance. PowerSlider does so with a new Flex SLO contract that turns bounded user slack into an optimization constraint, prefill--think--answer disaggregation exposing per-stage frequency and KV control, and a Karush--Kuhn--Tucker (KKT) online solver re-solving within 7.7 ms of every cap change, backed by a consolidated fail-safe that power-gates drained instances when DVFS bottoms out on static power. On SGLang with production traces, sys sustains 78.3% online goodput at a 30% cap reduction versus 47.6% for the best of five baselines (1.64×), holds latency-critical tails within 1.3× of nominal (baselines: 2.3--6×, up to 12×), and delivers 92% mean goodput through a replayed CAISO grid-emergency day bottoming at 0.41× (54% at the trough; every baseline below 7%).