G-TADOC: Enabling Efficient GPU-Based Text Analytics without Decompression

TL;DR

G-TADOC is a novel GPU framework that enables efficient text analytics directly on compressed data without decompression, overcoming dependency, synchronization, and sequence maintenance challenges to significantly accelerate processing.

Contribution

It introduces a GPU-based framework for direct text analytics on compressed data, with innovative workload scheduling, thread-safe memory management, and sequence preservation strategies.

Findings

Achieves 31.1x average speedup over state-of-the-art TADOC.

First GPU framework for direct text analytics on compressed data.

Effectively handles dependencies, synchronization, and sequence maintenance.

Abstract

Text analytics directly on compression (TADOC) has proven to be a promising technology for big data analytics. GPUs are extremely popular accelerators for data analytics systems. Unfortunately, no work so far shows how to utilize GPUs to accelerate TADOC. We describe G-TADOC, the first framework that provides GPU-based text analytics directly on compression, effectively enabling efficient text analytics on GPUs without decompressing the input data. G-TADOC solves three major challenges. First, TADOC involves a large amount of dependencies, which makes it difficult to exploit massive parallelism on a GPU. We develop a novel fine-grained thread-level workload scheduling strategy for GPU threads, which partitions heavily-dependent loads adaptively in a fine-grained manner. Second, in developing G-TADOC, thousands of GPU threads writing to the same result buffer leads to inconsistency while directly using locks and atomic operations lead to large synchronization overheads. We develop a memory pool with thread-safe data structures on GPUs to handle such difficulties. Third, maintaining the sequence information among words is essential for lossless compression. We design a sequence-support strategy, which maintains high GPU parallelism while ensuring sequence information. Our experimental evaluations show that G-TADOC provides 31.1x average speedup compared to state-of-the-art TADOC.