A New Data Layout for Set Intersection on GPUs

Abstract

Abstract-Set intersection is the core in a variety of problems, e.g. frequent itemset mining and sparse boolean matrix multiplication. It is well-known that large speed gains can, for some computational problems, be obtained by using a graphics processing unit (GPU) as a massively parallel computing device. However, GPUs require highly regular control flow and memory access patterns, and for this reason previous GPU methods for intersecting sets have used a simple bitmap representation. This representation requires excessive space on sparse data sets. In this paper we present a novel data layout, BATMAP, that is particularly well suited for parallel processing, and is compact even for sparse data.Frequent itemset mining is one of the most important applications of set intersection. As a case-study on the potential of BATMAPs we focus on frequent pair mining, which is a core special case of frequent itemset mining. The main finding is that our method is able to achieve speedups over both Apriori and FP-growth when the number of distinct items is large, and the density of the problem instance is above 1%. Previous implementations of frequent itemset mining on GPU have not been able to show speedups over the best single-threaded implementations.

Answering Spatial Multiple-Set Intersection Queries Using 2-3 Cuckoo Hash-Filters

“…It is known that a load strictly less than 1/6 allows all items to be placed with high probability, but a load strictly greater than 1/6 will fail to place all items with high probability. These results are discussed and proven by Amossen and Pagh [2] and Loh and Pagh [31] (see also the related combinatorial results in [24]).…”

Probabilistic Frequent Itemset Mining on a GPU Cluster

“…Frequent itemset mining on GPUs has been also studied [5], [7], [14], [16]. Fang et al [7] proposed two approaches, GPU-based and CPU-GPU hybrid methods.…”

“…Teodoro et al [16] parallelized the Tree Projection algorithm [1] on a GPU as well as a multicore CPU. Amossen et al [5] presented a novel data layout BatMap to represent bitstrings. This layout is well suited to parallel processing and compact even for sparse data.…”

“…GPGPU refers to performing computation on GPUs (Graphics Pro- cessing Units), which are originally designed for processing 3D graphics. GPGPU has received much attention from not only the field of high performance computing but also many other fields such as data mining [5], [7], [12], [14], [16]. This is because GPUs have more than hundred processing units and can process many data elements with high parallelism.…”

A Highly Parallel Reuse Distance Analysis Algorithm on GPUs

“…al [32] presented a hash table implementation on GPU both with and without atomic operations supported. Amossen and Pagh [3] introduced a new data layout, BATMAP, to accelerate item set mining for set intersections.…”