arXiv subjects

Eishi Arima

Publications and source records attributed to Eishi Arima.

7 recordsLinked to original sources

arXiv2026

CEDD-optimizer: Enabling Cost-Efficient Dataset Distillation on Geographically Distributed Edge Systems

Centralized learning is a fundamental paradigm in modern AI, in which data are collected from distributed edge devices and aggregated at a central host for model training. However, this pipeline is often bottlenecked by the substantial communication overhead of data collection. Dataset Distillation (DD), with its high compression ratio, is therefore attractive for centralized learning on distributed data. Yet, the cost efficiency of DD in non-uniform edge environments remains largely unexplored. Geographically distributed edge devices face different energy and data-transfer prices, while key DD hyperparameters, such as the target compression ratio and number of distillation steps, substantially affect energy use, transfer overhead, and downstream test accuracy. These hyperparameters must therefore be tuned both locally and globally. We propose the Cost-Efficient Dataset Distillation optimizer (CEDD-optimizer), a hyperparameter-tuning framework for cost-efficient distributed DD. It minimizes total cost under a training-quality constraint by optimizing hyperparameter settings across edge devices while accounting for environmental non-uniformity. The framework comprises two modules: CEDD-calibrator and CEDD-solver. CEDD-calibrator identifies parameters for our energy and training-quality models: the former is determined through offline calibration, whereas the latter is estimated online through a three-step tuning scheme. Based on these models, CEDD-solver solves the cost-minimization problem to steer and improve the distributed DD workflow. Experiments across various image datasets show that our approach achieves up to a 20.8 improvement over the baseline DD method under the same quality constraint.

cs.DC↗

arXiv2024

Hierarchical Resource Partitioning on Modern GPUs: A Reinforcement Learning Approach

GPU-based heterogeneous architectures are now commonly used in HPC clusters. Due to their architectural simplicity specialized for data-level parallelism, GPUs can offer much higher computational throughput and memory bandwidth than CPUs in the same generation do. However, as the available resources in GPUs have increased exponentially over the past decades, it has become increasingly difficult for a single program to fully utilize them. As a consequence, the industry has started supporting several resource partitioning features in order to improve the resource utilization by co-scheduling multiple programs on the same GPU die at the same time. Driven by the technological trend, this paper focuses on hierarchical resource partitioning on modern GPUs, and as an example, we utilize a combination of two different features available on recent NVIDIA GPUs in a hierarchical manner: MPS (Multi-Process Service), a finer-grained logical partitioning; and MIG (Multi-Instance GPU), a coarse-grained physical partitioning. We propose a method for comprehensively co-optimizing the setup of hierarchical partitioning and the selection of co-scheduling groups from a given set of jobs, based on reinforcement learning using their profiles. Our thorough experimental results demonstrate that our approach can successfully set up job concurrency, partitioning, and co-scheduling group selections simultaneously. This results in a maximum throughput improvement by a factor of 1.87 compared to the time-sharing scheduling.

cs.DC↗

arXiv2024

Orchestrated Co-scheduling, Resource Partitioning, and Power Capping on CPU-GPU Heterogeneous Systems via Machine Learning

CPU-GPU heterogeneous architectures are now commonly used in a wide variety of computing systems from mobile devices to supercomputers. Maximizing the throughput for multi-programmed workloads on such systems is indispensable as one single program typically cannot fully exploit all available resources. At the same time, power consumption is a key issue and often requires optimizing power allocations to the CPU and GPU while enforcing a total power constraint, in particular when the power/thermal requirements are strict. The result is a system-wide optimization problem with several knobs. In particular we focus on (1) co-scheduling decisions, i.e., selecting programs to co-locate in a space sharing manner; (2) resource partitioning on both CPUs and GPUs; and (3) power capping on both CPUs and GPUs. We solve this problem using predictive performance modeling using machine learning in order to coordinately optimize the above knob setups. Our experiential results using a real system show that our approach achieves up to 67% of speedup compared to a time-sharing-based scheduling with a naive power capping that evenly distributes power budgets across components.

cs.DC↗

arXiv2024

On the Convergence of Malleability and the HPC PowerStack: Exploiting Dynamism in Over-Provisioned and Power-Constrained HPC Systems

Recent High-Performance Computing (HPC) systems are facing important challenges, such as massive power consumption, while at the same time significantly under-utilized system resources. Given the power consumption trends, future systems will be deployed in an over-provisioned manner where more resources are installed than they can afford to power simultaneously. In such a scenario, maximizing resource utilization and energy efficiency, while keeping a given power constraint, is pivotal. Driven by this observation, in this position paper we first highlight the recent trends of resource management techniques, with a particular focus on malleability support (i.e., dynamically scaling resource allocations/requirements for a job), co-scheduling (i.e., co-locating multiple jobs within a node), and power management. Second, we consider putting them together, assess their relationships/synergies, and discuss the functionality requirements in each software component for future over-provisioned and power-constrained HPC systems. Third, we briefly introduce our ongoing efforts on the integration of software tools, which will ultimately lead to the convergence of malleability and power management, as it is designed in the HPC PowerStack initiative.

cs.DC↗

arXiv2024

Optimizing Hardware Resource Partitioning and Job Allocations on Modern GPUs under Power Caps

CPU-GPU heterogeneous systems are now commonly used in HPC (High-Performance Computing). However, improving the utilization and energy-efficiency of such systems is still one of the most critical issues. As one single program typically cannot fully utilize all resources within a node/chip, co-scheduling (or co-locating) multiple programs with complementary resource requirements is a promising solution. Meanwhile, as power consumption has become the first-class design constraint for HPC systems, such co-scheduling techniques should be well-tailored for power-constrained environments. To this end, the industry recently started supporting hardware-level resource partitioning features on modern GPUs for realizing efficient co-scheduling, which can operate with existing power capping features. For example, NVidia's MIG (Multi-Instance GPU) partitions one single GPU into multiple instances at the granularity of a GPC (Graphics Processing Cluster). In this paper, we explicitly target the combination of hardware-level GPU partitioning features and power capping for power-constrained HPC systems. We provide a systematic methodology to optimize the combination of chip partitioning, job allocations, as well as power capping based on our scalability/interference modeling while taking a variety of aspects into account, such as compute/memory intensity and utilization in heterogeneous computational resources (e.g., Tensor Cores). The experimental result indicates that our approach is successful in selecting a near optimal combination across multiple different workloads.

cs.DC↗

arXiv2023

GreenCourier: Carbon-Aware Scheduling for Serverless Functions

This paper presents GreenCourier, a novel scheduling framework that enables the runtime scheduling of serverless functions across geographically distributed regions based on their carbon efficiencies. Our framework incorporates an intelligent scheduling strategy for Kubernetes and supports Knative as the serverless platform. To obtain real-time carbon information for different geographical regions, our framework supports multiple marginal carbon emissions sources such as WattTime and the Carbon-aware SDK. We comprehensively evaluate the performance of our framework using the Google Kubernetes Engine and production serverless function traces for scheduling functions across Spain, France, Belgium, and the Netherlands. Results from our experiments show that compared to other approaches, GreenCourier reduces carbon emissions per function invocation by an average of 13.25%.

cs.DC↗

arXiv2023

Sustainability in HPC: Vision and Opportunities

Tackling climate change by reducing and eventually eliminating carbon emissions is a significant milestone on the path toward establishing an environmentally sustainable society. As we transition into the exascale era, marked by an increasing demand and scale of HPC resources, the HPC community must embrace the challenge of reducing carbon emissions from designing and operating modern HPC systems. In this position paper, we describe challenges and highlight different opportunities that can aid HPC sites in reducing the carbon footprint of modern HPC systems.

cs.DC↗