Saved in:
Bibliographic Details
Main Authors: Mileski, Dimitar, Petrovski, Nikola, Gusev, Marjan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.21033
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913761003044864
author Mileski, Dimitar
Petrovski, Nikola
Gusev, Marjan
author_facet Mileski, Dimitar
Petrovski, Nikola
Gusev, Marjan
contents Training large language models requires extensive processing, made possible by many high-performance computing resources. This study compares multi-node and multi-GPU environments for training large language models of electrocardiograms. It provides a detailed mapping of current frameworks for distributed deep learning in multinode and multi-GPU settings, including Horovod from Uber, DeepSpeed from Microsoft, and the built-in distributed capabilities of PyTorch and TensorFlow. We compare various multi-GPU setups for different dataset configurations, utilizing multiple HPC nodes independently and focusing on scalability, speedup, efficiency, and overhead. The analysis leverages HPC infrastructure with SLURM, Apptainer (Singularity) containers, CUDA, PyTorch, and shell scripts to support training workflows and automation. We achieved a sub-linear speedup when scaling the number of GPUs, with values of 1.6x for two and 1.9x for four.
format Preprint
id arxiv_https___arxiv_org_abs_2503_21033
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scalability Evaluation of HPC Multi-GPU Training for ECG-based LLMs
Mileski, Dimitar
Petrovski, Nikola
Gusev, Marjan
Distributed, Parallel, and Cluster Computing
C.4; I.6
Training large language models requires extensive processing, made possible by many high-performance computing resources. This study compares multi-node and multi-GPU environments for training large language models of electrocardiograms. It provides a detailed mapping of current frameworks for distributed deep learning in multinode and multi-GPU settings, including Horovod from Uber, DeepSpeed from Microsoft, and the built-in distributed capabilities of PyTorch and TensorFlow. We compare various multi-GPU setups for different dataset configurations, utilizing multiple HPC nodes independently and focusing on scalability, speedup, efficiency, and overhead. The analysis leverages HPC infrastructure with SLURM, Apptainer (Singularity) containers, CUDA, PyTorch, and shell scripts to support training workflows and automation. We achieved a sub-linear speedup when scaling the number of GPUs, with values of 1.6x for two and 1.9x for four.
title Scalability Evaluation of HPC Multi-GPU Training for ECG-based LLMs
topic Distributed, Parallel, and Cluster Computing
C.4; I.6
url https://arxiv.org/abs/2503.21033