AMPED: Accelerating MTTKRP for Billion-Scale Sparse Tensor Decomposition on Multiple GPUs

TL;DR

AMPED is a multi-GPU algorithm that significantly accelerates MTTKRP computations for billion-scale sparse tensors, enabling large-scale tensor decomposition with improved speed and resource utilization.

Contribution

The paper introduces AMPED, a novel multi-GPU parallel algorithm with a partitioning and load balancing strategy for efficient large-scale sparse tensor decomposition.

Findings

Achieves 5.1x speedup over state-of-the-art GPU methods.

Scales beyond single GPU memory and performance limits.

Effectively handles billion-scale sparse tensors.

Abstract

Matricized Tensor Times Khatri-Rao Product (MTTKRP) is the computational bottleneck in sparse tensor decomposition. As real-world sparse tensors grow to billions of nonzeros, they increasingly demand higher memory capacity and compute throughput from hardware accelerators. In this work, we present AMPED, a multi-GPU parallel algorithm designed to accelerate MTTKRP on billion-scale sparse tensors. AMPED scales beyond the limits of a single GPU, meeting both the memory and performance requirements of large-scale workloads. We introduce a partitioning strategy combined with a dynamic load balancing scheme to distribute computation and minimize GPU idle time. On real-world billion-scale tensors, AMPED achieves a 5.1x geometric mean speedup in total execution time over state-of-the-art GPU baselines using 4 GPUs on a single CPU node.