paper-with-me

Papers

HetCCL: Accelerating LLM Training with Heterogeneous GPUs

2026-01-30 · Heehoon Kim, Jaehwan Lee, Taejeoung Kim, Jongwon Park, Jinpyo Kim, Pyongwon Suh, Ryan H. Choi, Sangwoo Lee, Jaejin Lee arxiv

The rapid growth of large language models is driving organizations to expand their GPU clusters, often with GPUs from multiple vendors. However, current deep learning frameworks lack support for collective communication across heterogeneous GPUs, leading to inefficiency and higher costs. We present HetCCL, a collective communication library that unifies vendor-specific backends and enables RDMA-based communication across GPUs without requiring driver modifications. HetCCL introduces two novel mechanisms that enable cross-vendor communication while leveraging optimized vendor libraries, NVIDIA NCCL and AMD RCCL. Evaluations on a multi-vendor GPU cluster show that HetCCL matches NCCL and RCCL performance in homogeneous setups while uniquely scaling in heterogeneous environments, enabling practical, high-performance training with both NVIDIA and AMD GPUs without changes to existing deep learning applications.

📄 PDF Abstract BibTeX arXiv:2601.22585

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

HetCCL: Enabling Collective Communication For Mixed-Vendor Heterogeneous Clusters

2026-05-29 · Yuejie Wang, Tao Chang, Yuanyuan Zhao, Yulong Ao 외 arxiv

Training Large Language Models (LLMs) on heterogeneous clusters presents significant challenges for collective communication, as hardware from multiple vendors introduces diverse network and computational characteristics…

Large Scale Artificial Neural Network Training Using Multi-GPUs

2015-11-13 · Linnan Wang, Wei Wu, Jianxiong Xiao, Yang Yi

This paper describes a method for accelerating large scale Artificial Neural Networks (ANN) training using multi-GPUs by reducing the forward and backward passes to matrix multiplication. We propose an out-of-core multi-…

GPU

Heterogeneous Decentralized Diffusion Models

2026-03-06 · Zhiying Jiang, Raihan Seraj, Marcos Villagra, Bidhan Roy arxiv

Training frontier-scale diffusion models often requires substantial computational resources concentrated in tightly-coupled clusters, limiting participation to well-resourced institutions. While Decentralized Diffusion M…

Optimus: Accelerating Large-Scale Multi-Modal LLM Training by Bubble Exploitation

2024-08-07 · Weiqi Feng, Yangrui Chen, Shaoyu Wang, Yanghua Peng 외

Multimodal large language models (MLLMs) have extended the success of large language models (LLMs) to multiple data types, such as image, text and audio, achieving significant performance in various domains, including mu…

GPUQuestion AnsweringSchedulingVisual Question Answering

Astra: Efficient and Money-saving Automatic Parallel Strategies Search on Heterogeneous GPUs

2025-02-19 · Peiran Wang, Haibing Li, Fu Haohan, Shiyong Li 외

In this paper, we introduce an efficient and money-saving automatic parallel strategies search framework on heterogeneous GPUs: Astra. First, Astra searches for the efficiency-optimal parallel strategy in both GPU config…

GPU