Efficient and high‐quality sparse graph coloring on GPUs

Abstract

Summary Graph coloring has been broadly used to discover concurrency in parallel computing. To speed up graph coloring for large‐scale datasets, parallel algorithms have been proposed to leverage modern GPUs. Existing GPU implementations either have limited performance or yield unsatisfactory coloring quality (too many colors assigned). We present a work‐efficient parallel graph coloring implementation on GPUs with good coloring quality. Our approach uses the speculative greedy scheme, which inherently yields better quality than the method of finding maximal independent set. To achieve high performance on GPUs, we refine the algorithm to leverage efficient operators and alleviate conflicts. We also incorporate common optimization techniques to further improve performance. Our method is evaluated with both synthetic and real‐world sparse graphs on the NVIDIA GPU. Experimental results show that our proposed implementation achieves averaged 4.1 × (up to 8.9 × ) speedup over the serial implementation. It also outperforms the existing GPU implementation from the NVIDIA CUSPARSE library (2.2 × average speedup), while yielding much better coloring quality than CUSPARSE.

HPGraph: High-Performance Graph Analytics with Productivity on the GPU

“…Graph analytics have been widely applied in many applications [39,40]. In this paper, we have presented HPGraph, a GPU graph analytics framework which maps a vertex programming to an optimized matrix backend.…”

Feluca: A Two-Stage Graph Coloring Algorithm With Color-Centric Paradigm on GPU

“…The experimental results show that Feluca achieves up to 8.39×, 14.70×, 7.55×, and 9.70× speed up over kokkos [20], Gunrock [36], SIRG [44] and ChenGC [42], [43], respectively. Table 4 shows that Feluca outperforms all other competitors in terms of run-time with all ten datasets.…”

“…We compared Feluca with some state-of-the-art methods in this area, such as kokkos [20], Gunrock [36], GraphBLAST [41], ChenGC [42], [43], SIRG [44], cuSPARSE [40] and JPL [40]. In this experiment, Feluca switches the execution stage by setting α to 10%.…”

“…SIRG [44] proposes a re-coloring scheme, by using a data-driven programming method, to handle the conflict vertices. Chen et al [42], [43] assign the threads by using a work-efficient manner, which is similar with Kokkos [20]. But the authors use a greedy algorithm to assign the colors.…”

clMF: A fine-grained and portable alternating least squares algorithm for parallel matrix factorization

“…Compared to stochastic gradient descent (SGD) [8,9], the ALS algorithm is not only inherently parallel, but can incorporate implicit ratings [1]. Nevertheless, the ALS algorithm involves parallel sparse matrix manipulation [10] which is challenging to achieve high performance due to imbalanced workload [11,12,13], random memory access [14,15], unpredictable amount of computations [16] and task dependency [17,18,19]. This particularly holds when parallelizing and optimizing ALS on modern multi-cores and many-cores [20].…”