CORTEXA
← Browse
arxivcs.LG2026-07-17

PagedWeight: Efficient MoE LLM Serving with Dynamic Quality-Aware Weight Quantization

Yuchen Yang, Yifan Zhao, Anisha Dasgupta, Sasa Misailovic

Mixture-of-Experts (MoE) is a popular class of large language models (LLMs), offering high efficiency and accuracy. However, in KV-cache-intensive serving scenarios, MoEs often exhibit a tension between the GPU memory requirements of the model weights and the growing KV cache. We propose PagedWeight, a novel management method for MoE LLM serving that dynamically quantizes MoE model's weights at runtime and balances expert-weight precision with the KV cache sizes. PagedWeight exposes and effectively navigates the complex tradeoff between the model's task accuracy, memory consumption, and throughput/latency. Across several memory-sensitive MoE serving scenarios, PagedWeight improves the quality-memory tradeoff over several existing quantization baselines. PagedWeight achieves FP16-equivalent accuracy with up to 72.0% GPU memory savings and 1.94$\times$ throughput improvement, and improves quality over quantization methods by up to 39.3% at a similar memory budget with at most 4.1% throughput loss.

View free PDFSource page

Related papers

arxivcs.LGcs.AI2026-07-20

MXSens: Sensitivity-Aware Mixed-Precision Quantization for Efficient LLM Inference

Simla Burcu Harma, Danila Mishin, Zhengyuan Su, Ayan Chakraborty, Elizaveta Kostenok, Dongho Ha, et al.

4-bit quantization enables efficient LLM inference, but suffers from significant accuracy degradation due to outliers. Prior work addresses this problem via data rotation or mixed-precision integer quantization, but often relies on software-managed scaling and frequent dequantiza…

View free PDFSource page
arxivcs.ARcs.AIcs.ETcs.LG2026-07-29

LLMET: Enabling Cross-Layer Evaluation of Emerging M3D Memories for Energy-Efficient LLM Serving

Ming-Yen Lee, Hanchen Yang, Faaiq Waqar, Harsono Simka, Tushar Krishna, Muhammed Ahosan Ul Karim, et al.

The energy consumption of Large Language Model (LLM) serving is becoming a major system challenge as deployment scales, driven by hardware power and thermal constraints and rising electricity costs. A key contributor to chip energy dissipation is data movement between limited on-…

View free PDFSource page
arxivcs.LG2026-07-01

Beyond Activation Alignment:The Alignment-Diversity Tradeoff in Task-Aware LLM Quantization

Fei Wang, Chao Xue, Taoran Liu, Li Shen, Ye Liu, ChangXing Ding

Mixed-precision quantization (MPQ) has become a key technique for deploying large language models under stringent memory and compute constraints. We first identify a phenomenon that we term the Perplexity Illusion: layers ranked as important by perplexity-based sensitivity show l…

View free PDFSource page
arxivcs.LGcs.DC2026-07-01

MosaicKV: Serving Long-Context LLM with Dynamic Two-D KV Cache Compression

Sheng Qiang, Ruiwei Chen, Yinpeng Wu, Jinyu Gu, Zhichao Hua, Yubin Xia, et al.

Long-context LLM services now sustain prompts with hundreds of thousands to millions of tokens, making the key-value (KV) cache a first-order serving cost. Because the cache grows linearly with context length, it can exhaust GPU memory, force smaller batches, and reduce serving t…

View free PDFSource page