Efficient, Out-of-Memory Sparse MTTKRP on Massively Parallel Architectures

TL;DR

This paper introduces a novel GPU framework for tensor decomposition that efficiently handles out-of-memory data, significantly accelerating computations and outperforming existing methods on real-world sparse tensors.

Contribution

The paper proposes the BLCO format and adaptive strategies for out-of-memory tensor operations, enabling efficient, conflict-resolving parallel computations on GPUs.

Findings

Achieves 2.12-2.6X speedup over state-of-the-art methods.

Supports out-of-memory tensor processing on GPUs.

Reduces synchronization costs and improves in-memory performance.

Abstract

Tensor decomposition (TD) is an important method for extracting latent information from high-dimensional (multi-modal) sparse data. This study presents a novel framework for accelerating fundamental TD operations on massively parallel GPU architectures. In contrast to prior work, the proposed Blocked Linearized Coordinate (BLCO) format enables efficient out-of-memory computation of tensor algorithms using a unified implementation that works on a single tensor copy. Our adaptive blocking and linearization strategies not only meet the resource constraints of GPU devices, but also accelerate data indexing, eliminate control-flow and memory-access irregularities, and reduce kernel launching overhead. To address the substantial synchronization cost on GPUs, we introduce an opportunistic conflict resolution algorithm, in which threads collaborate instead of contending on memory access to discover and resolve their conflicting updates on-the-fly, without keeping any auxiliary information or storing non-zero elements in specific mode orientations. As a result, our framework delivers superior in-memory performance compared to prior state-of-the-art, and is the only framework capable of processing out-of-memory tensors. On the latest Intel and NVIDIA GPUs, BLCO achieves 2.12-2.6X geometric-mean speedup (with up to 33.35X speedup) over the state-of-the-art mixed-mode compressed sparse fiber (MM-CSF) on a range of real-world sparse tensors.