Saved in:
Bibliographic Details
Main Authors: Quach, Ryan, Wang, Yidi, Jahanshahi, Ali, Wong, Daniel, Kim, Hyoseung
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.12598
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909649236656128
author Quach, Ryan
Wang, Yidi
Jahanshahi, Ali
Wong, Daniel
Kim, Hyoseung
author_facet Quach, Ryan
Wang, Yidi
Jahanshahi, Ali
Wong, Daniel
Kim, Hyoseung
contents As AI inference becomes mainstream, research has begun to focus on improving the energy consumption of inference servers. Inference kernels commonly underutilize a GPU's compute resources and waste power from idling components. To improve utilization and energy efficiency, multiple models can co-locate and share the GPU. However, typical GPU spatial partitioning techniques often experience significant overheads when reconfiguring spatial partitions, which can waste additional energy through repartitioning overheads or non-optimal partition configurations. In this paper, we present ECLIP, a framework to enable low-overhead energy-efficient kernel-wise resource partitioning between co-located inference kernels. ECLIP minimizes repartitioning overheads by pre-allocating pools of CU masked streams and assigns optimal CU assignments to groups of kernels through our resource allocation optimizer. Overall, ECLIP achieves an average of 13% improvement to throughput and 25% improvement to energy efficiency.
format Preprint
id arxiv_https___arxiv_org_abs_2506_12598
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ECLIP: Energy-efficient and Practical Co-Location of ML Inference on Spatially Partitioned GPUs
Quach, Ryan
Wang, Yidi
Jahanshahi, Ali
Wong, Daniel
Kim, Hyoseung
Systems and Control
As AI inference becomes mainstream, research has begun to focus on improving the energy consumption of inference servers. Inference kernels commonly underutilize a GPU's compute resources and waste power from idling components. To improve utilization and energy efficiency, multiple models can co-locate and share the GPU. However, typical GPU spatial partitioning techniques often experience significant overheads when reconfiguring spatial partitions, which can waste additional energy through repartitioning overheads or non-optimal partition configurations. In this paper, we present ECLIP, a framework to enable low-overhead energy-efficient kernel-wise resource partitioning between co-located inference kernels. ECLIP minimizes repartitioning overheads by pre-allocating pools of CU masked streams and assigns optimal CU assignments to groups of kernels through our resource allocation optimizer. Overall, ECLIP achieves an average of 13% improvement to throughput and 25% improvement to energy efficiency.
title ECLIP: Energy-efficient and Practical Co-Location of ML Inference on Spatially Partitioned GPUs
topic Systems and Control
url https://arxiv.org/abs/2506.12598