arXiv subjects

Eun Kyung Lee

Publications and source records attributed to Eun Kyung Lee.

10 recordsLinked to original sources

arXiv2026

EnergyLens: Predictive Energy-Aware Exploration for Multi-GPU LLM Inference Optimization

We present EnergyLens, an end-to-end framework for energy-aware large language model (LLM) inference optimization. As LLMs scale, predicting and reducing their energy footprint has become critical for sustainability and datacenter operations, yet existing approaches either require production-level code and expensive profiling or fail to accurately capture multi-GPU energy behavior. As a result, practitioners lack tools for deciding which optimizations to prioritize and for selecting among existing deployment configurations when exhaustive profiling is impractical. EnergyLens addresses this gap with an intuitive einsum-based interface that captures LLM specifications including fusion, parallelism, and compute-communication overlap, combined with load-imbalance-aware MoE modeling and an empirically driven communication energy model for multi-GPU settings. We validate EnergyLens on Llama3 and Qwen3-MoE across tensor-parallel and expert-parallel configurations, achieving mean absolute percentage errors (MAPEs) between 9.25% and 13.19% for multi-GPU prefill and decode energy, and 12.97% across SM allocations for Megatron-style overlap. Our energy-driven exploration reveals up to 1.47x and 52.9x energy variation across configurations in prefill and decode efficiency and motivates distributed serving. We further show that compute-communication overlap is difficult to optimize with intuition alone, but EnergyLens correctly identifies Pareto-optimal overlap configurations.

cs.LG↗

arXiv2026

EnergAIzer: Fast and Accurate GPU Power Estimation Framework for AI Workloads

As AI workloads drive increases in datacenter power consumption, accurate GPU power estimation is critical for proactive power management. However, existing power models face a scalability bottleneck not in the modeling techniques themselves, but in obtaining the hardware utilization inputs they require. Conventional approaches rely on either costly simulation or hardware profiling, which makes them impractical when rapid predictions are required. This work presents EnergAIzer, which addresses this scalability bottleneck by developing a lightweight solution to predict utilization inputs, reducing the estimation walltime from hours to seconds. Our key insight is that kernels in AI workloads commonly employ optimizations that create structured patterns, which analytically determine memory traffic and execution timeline. We construct a performance model using these patterns as an analytical scaffold for empirical data fitting, which also naturally exposes module-level utilization. This predicted utilization is then fed into our power model to estimate dynamic power consumption. EnergAIzer achieves 8% power errors on NVIDIA Ampere GPUs, competitive with traditional power models with elaborate cycle-level simulation or hardware profiling. We demonstrate EnergAIzer's exploration capabilities for frequency scaling and architectural configurations, including forecasting the power of NVIDIA H100 with just 7% error. In summary, EnergAIzer provides fast and accurate power prediction for AI workloads, paving the way for power-aware design explorations.

cs.AR↗

arXiv2025

Revisiting Disaggregated Large Language Model Serving for Performance and Energy Implications

Different from traditional Large Language Model (LLM) serving that colocates the prefill and decode stages on the same GPU, disaggregated serving dedicates distinct GPUs to prefill and decode workload. Once the prefill GPU completes its task, the KV cache must be transferred to the decode GPU. While existing works have proposed various KV cache transfer paths across different memory and storage tiers, there remains a lack of systematic benchmarking that compares their performance and energy efficiency. Meanwhile, although optimization techniques such as KV cache reuse and frequency scaling have been utilized for disaggregated serving, their performance and energy implications have not been rigorously benchmarked. In this paper, we fill this research gap by re-evaluating prefill-decode disaggregation under different KV transfer mediums and optimization strategies. Specifically, we include a new colocated serving baseline and evaluate disaggregated setups under different KV cache transfer paths. Through GPU profiling using dynamic voltage and frequency scaling (DVFS), we identify and compare the performance-energy Pareto frontiers across all setups to evaluate the potential energy savings enabled by disaggregation. Our results show that performance benefits from prefill-decode disaggregation are not guaranteed and depend on the request load and KV transfer mediums. In addition, stage-wise independent frequency scaling enabled by disaggregation does not lead to energy saving due to inherently higher energy consumption of disaggregated serving.

cs.PF↗

arXiv2025

Refractive Index-Correlated Pseudocoloring for Adaptive Color Fusion in Holotomographic Cytology

Conventional bright-field (BF) cytology of thyroid fine-needle aspiration biopsy (FNAB) suffers from staining variability and limited subcellular contrast. Here, we present a refractive index-correlated pseudocoloring (RICP) framework that integrates quantitative refractive index (RI) maps obtained by holotomography (HT) with color BF images to enhance diagnostic interpretability. The imaging platform combines a digital micromirror device (DMD)-based HT system with an RGB LED illumination module, enabling simultaneous acquisition of RI tomograms and BF images from PAP-stained thyroid samples. The RICP algorithm adaptively embeds RI-derived structural information into the least-occupied hue channel, preserving color fidelity while enhancing nuclear and cytoplasmic contrast. Applied to benign and malignant thyroid clusters, RICP revealed diagnostically relevant features such as nucleoli, lipid droplets, and nuclear irregularities, and hue-saturation analysis quantitatively differentiated cytological categories. This perceptually grounded, label-free framework bridges conventional color cytology and quantitative optical imaging for improved diagnostic precision.

physics.optics↗

arXiv2025

Towards Pareto Optimal Throughput in Small Language Model Serving

Large language models (LLMs) have revolutionized the state-of-the-art of many different natural language processing tasks. Although serving LLMs is computationally and memory demanding, the rise of Small Language Models (SLMs) offers new opportunities for resource-constrained users, who now are able to serve small models with cutting-edge performance. In this paper, we present a set of experiments designed to benchmark SLM inference at performance and energy levels. Our analysis provides a new perspective in serving, highlighting that the small memory footprint of SLMs allows for reaching the Pareto-optimal throughput within the resource capacity of a single accelerator. In this regard, we present an initial set of findings demonstrating how model replication can effectively improve resource utilization for serving SLMs.

cs.CL↗

arXiv2025

Mind the Memory Gap: Unveiling GPU Bottlenecks in Large-Batch LLM Inference

Large language models have been widely adopted across different tasks, but their auto-regressive generation nature often leads to inefficient resource utilization during inference. While batching is commonly used to increase throughput, performance gains plateau beyond a certain batch size, especially with smaller models, a phenomenon that existing literature typically explains as a shift to the compute-bound regime. In this paper, through an in-depth GPU-level analysis, we reveal that large-batch inference remains memory-bound, with most GPU compute capabilities underutilized due to DRAM bandwidth saturation as the primary bottleneck. To address this, we propose a Batching Configuration Advisor (BCA) that optimizes memory allocation, reducing GPU memory requirements with minimal impact on throughput. The freed memory and underutilized GPU compute capabilities can then be leveraged by concurrent workloads. Specifically, we use model replication to improve serving throughput and GPU utilization. Our findings challenge conventional assumptions about LLM inference, offering new insights and practical strategies for improving resource utilization, particularly for smaller language models. The code is publicly available at https://github.com/FerranAgulloLopez/vLLMBatchingMemoryGap.

cs.DC↗

arXiv2025

Towards Efficient Key-Value Cache Management for Prefix Prefilling in LLM Inference

The increasing adoption of large language models (LLMs) with extended context windows necessitates efficient Key-Value Cache (KVC) management to optimize inference performance. Inference workloads like Retrieval-Augmented Generation (RAG) and agents exhibit high cache reusability, making efficient caching critical to reducing redundancy and improving speed. We analyze real-world KVC access patterns using publicly available traces and evaluate commercial key-value stores like Redis and state-of-the-art RDMA-based systems (CHIME [1] and Sherman [2]) for KVC metadata management. Our work demonstrates the lack of tailored storage solution for KVC prefilling, underscores the need for an efficient distributed caching system with optimized metadata management for LLM workloads, and provides insights into designing improved KVC management systems for scalable, low-latency inference.

cs.ET↗

arXiv2024

A Robust Power Model Training Framework for Cloud Native Runtime Energy Metric Exporter

Estimating power consumption in modern Cloud environments is essential for carbon quantification toward green computing. Specifically, it is important to properly account for the power consumed by each of the running applications, which are packaged as containers. This paper examines multiple challenges associated with this goal. The first challenge is that multiple customers are sharing the same hardware platform (multi-tenancy), where information on the physical servers is mostly obscured. The second challenge is the overhead in power consumption that the Cloud platform control plane induces. This paper addresses these challenges and introduces a novel pipeline framework for power model training. This allows versatile power consumption approximation of individual containers on the basis of available performance counters and other metrics. The proposed model utilizes machine learning techniques to predict the power consumed by the control plane and associated processes, and uses it for isolating the power consumed by the user containers, from the server power consumption. To determine how well the prediction results in an isolation, we introduce a metric termed isolation goodness. Applying the proposed power model does not require online power measurements, nor does it need information on the physical servers, configuration, or information on other tenants sharing the same machine. The results of cross-workload, cross-platform experiments demonstrated the higher accuracy of the proposed model when predicting power consumption of unseen containers on unknown platforms, including on virtual machines.

cs.DC↗

arXiv2024

FedCore: Straggler-Free Federated Learning with Distributed Coresets

Federated learning (FL) is a machine learning paradigm that allows multiple clients to collaboratively train a shared model while keeping their data on-premise. However, the straggler issue, due to slow clients, often hinders the efficiency and scalability of FL. This paper presents FedCore, an algorithm that innovatively tackles the straggler problem via the decentralized selection of coresets, representative subsets of a dataset. Contrary to existing centralized coreset methods, FedCore creates coresets directly on each client in a distributed manner, ensuring privacy preservation in FL. FedCore translates the coreset optimization problem into a more tractable k-medoids clustering problem and operates distributedly on each client. Theoretical analysis confirms FedCore's convergence, and practical evaluations demonstrate an 8x reduction in FL training time, without compromising model accuracy. Our extensive evaluations also show that FedCore generalizes well to existing FL frameworks.

cs.DC↗

arXiv2020

Measurement of the Background Activities of a 100Mo-enriched powder sample for AMoRE crystal material using a single high purity germanium detector

The Advanced Molybdenum-based Rare process Experiment (AMoRE) searches for neutrino-less double-beta (0ν\b{eta}\b{eta}) decay of 100Mo in enriched molybdate crystals. The AMoRE crystals must have low levels of radioactive contamination to achieve low background signals with energies near the Q-value of the 100Mo 0ν\b{eta}\b{eta} decay. To produce low-activity crystals, radioactive contaminants in the raw materials used to form the crystals must be controlled and quantified. 100EnrMoO3 powder, which is enriched in the 100Mo isotope, is of particular interest as it is the source of 100Mo in the crystals. A high-purity germanium detector having 100% relative efficiency, named CC1, is being operated in the Yangyang underground laboratory. Using CC1, we collected a gamma spectrum from a 1.6-kg 100EnrMoO3 powder sample enriched to 96.4% in 100Mo. Activities were analyzed for the isotopes 228Ac, 228Th, 226Ra, and 40K. They are long-lived naturally occurring isotopes that can produce background signals in the region of interest for AMoRE. Activities of both 228Ac and 228Th were < 1.0 mBq/kg at 90% confidence level (C.L.). The activity of 226Ra was measured to be 5.1 \pm 0.4 (stat) \pm 2.2 (syst) mBq/kg. The 40K activity was found as < 16.4 mBq/kg at 90% C.L.

physics.ins-det↗