Research

Paper

AI LLM March 17, 2026

An Efficient Heterogeneous Co-Design for Fine-Tuning on a Single GPU

Authors

Ruijia Yang, Zeyi Wen

Abstract

Fine-tuning Large Language Models (LLMs) has become essential for domain adaptation, but its memory-intensive property exceeds the capabilities of most GPUs. To address this challenge and democratize LLM fine-tuning, we present SlideFormer, a novel system designed for single-GPU environments. Our innovations are: (1) A lightweight asynchronous engine that treats the GPU as a sliding window and overlaps GPU computation with CPU updates and multi-tier I/O. (2) A highly efficient heterogeneous memory management scheme significantly reduces peak memory usage. (3) Optimized Triton kernels to solve key bottlenecks and integrated advanced I/O. This collaborative design enables fine-tuning of the latest 123B+ models on a single RTX 4090, supporting up to 8x larger batch sizes and 6x larger models. In evaluations, SlideFormer achieves 1.40x to 6.27x higher throughput while roughly halving CPU/GPU memory usage compared to baselines, sustaining >95% peak performance on both NVIDIA and AMD GPUs.

Metadata

arXiv ID: 2603.16428
Provider: ARXIV
Primary Category: cs.DC
Published: 2026-03-17
Fetched: 2026-03-18 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.16428v1</id>\n    <title>An Efficient Heterogeneous Co-Design for Fine-Tuning on a Single GPU</title>\n    <updated>2026-03-17T12:05:17Z</updated>\n    <link href='https://arxiv.org/abs/2603.16428v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.16428v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Fine-tuning Large Language Models (LLMs) has become essential for domain adaptation, but its memory-intensive property exceeds the capabilities of most GPUs. To address this challenge and democratize LLM fine-tuning, we present SlideFormer, a novel system designed for single-GPU environments. Our innovations are: (1) A lightweight asynchronous engine that treats the GPU as a sliding window and overlaps GPU computation with CPU updates and multi-tier I/O. (2) A highly efficient heterogeneous memory management scheme significantly reduces peak memory usage. (3) Optimized Triton kernels to solve key bottlenecks and integrated advanced I/O. This collaborative design enables fine-tuning of the latest 123B+ models on a single RTX 4090, supporting up to 8x larger batch sizes and 6x larger models. In evaluations, SlideFormer achieves 1.40x to 6.27x higher throughput while roughly halving CPU/GPU memory usage compared to baselines, sustaining &gt;95% peak performance on both NVIDIA and AMD GPUs.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.DC'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-03-17T12:05:17Z</published>\n    <arxiv:comment>7 pages</arxiv:comment>\n    <arxiv:primary_category term='cs.DC'/>\n    <author>\n      <name>Ruijia Yang</name>\n    </author>\n    <author>\n      <name>Zeyi Wen</name>\n    </author>\n  </entry>"
}