Search

Sparse general matrix multiplication (SpGEMM) is an important and expensive computation primitive in many real-world applications. Due to SpGEMM’s inherent irregularity and the vast diversity of its input matrices, developing high-performance SpGEMM implementation on modern processors such as GPUs is challenging. The state-of-the-art SpGEMM libraries (i.e., $nsparse$ and $spECK$ ) adopt several algorithms to tackle the challenges of global load balance, local load balance, and allocation of the result matrix. While these libraries focus on the high-level algorithm design for SpGEMM, they neglect several low-level architecture-specific optimizations, which causes inefficient implementations in their libraries. In this paper, we classify their inefficient implementations into several categories. Based on our observations, we propose a highly optimized SpGEMM library called $OpSparse$ . The optimizations in $OpSparse$ include 1) optimizing the binning method by improving the utilization of the shared memory, 2) optimizing the hashing method by reducing the accesses to the hash tables, 3) improving the trade-off between hash collision rate and hardware utilization in the hashing method by setting appropriate binning ranges, 4) reducing the overheads of global memory utilization by minimizing the global memory usage of the metadata, and 5) improving the execution parallelism by overlapping global memory allocation with kernel execution. Performance evaluations with 26 commonly used matrices on an Nvidia Tesla V100 GPU show that $OpSparse$ achieves on average $7.35\times $ (up to $27.8\times $ ), $1.43\times $ (up to $1.81\times $ ), and $1.52\times $ (up to $2.04\times $ ) speedups over three state-of-the-art SpGEMM libraries: $\mathit {cuSPARSE}$ , $nsparse$ , and $spECK$ , respectively.

Graph Neural Networks have drawn tremendous attention in the past few years due to their convincing performance and high interpretability in various graph-based tasks like link prediction and node classification. With the ever-growing graph size in the real world, especially for industrial graphs at a billion-level, the storage of graphs can easily consume Terabytes so that the process of GNNs has to be processed in a distributed manner. As a result, the execution could be inefficient due to the expensive cross-node communication and irregular memory access. Various GNN accelerators have been proposed for efficient GNN processing. They, however, mainly focused on small and medium-size graphs, which is not applicable to large-scale distributed graphs. In this paper, we present a practical Near-Data-Processing architecture based on a memory-pool system for large-scale distributed GNNs. We propose a customized memory fabric interface to construct the memory pool for low-latency and high throughput cross-node communication, which can provide flexible memory allocation and strong scalability. A practical Near-Data-Processing design is proposed for efficient work offloading and bandwidth utilization improvement. Moreover, we also introduce a partition and scheduling scheme to further improve performance and achieve workload balance. Comprehensive evaluations demonstrate that the proposed architecture can achieve up to $27\times $ and $8\times $ higher training speed compared to two state-of-the-art distributed GNN frameworks: Deep Graph Library and $P^{3}$ , respectively.

Raspite PbWO4, which is rare in nature, poses a great challenge to laboratory preparation and only limited optical and spectroscopic investigations have been conducted on raspite PbWO4crystals from nature. In this article, we report that raspite PbWO4nanobelts can be obtained by a facile composite-salt-mediated synthesis method without using any surface-capping agent. The nanobelts are well crystallized with lengths up to hundreds of microns. The synthesis method is simple and cost effective. Our investigation demonstrates that the raspite PbWO4nanobelts possess excellent blue–green luminescent property at low temperature and highly sensitive photocurrent response to light switches at room temperature. The growth mechanism of the raspite PbWO4nanobelts, the structure characteristic and the luminescent mechanism are discussed in detail. [ABSTRACT FROM AUTHOR]

SiCN nanowires are prepared by pyrolysis of hexamethyldisilazane using ferrocene as a catalyst precursor with or without thiophene, at 1100 °C in a flowing Ar atmosphere. The SiCN nanowires are about 10−60 nm in diameter and on the millimeter scale in length. The addition of thiophene is found to be effective in promoting the growth of SiCN nanowires. A vapor−liquid−solid growth mechanism of the nanowires is proposed. Electrical resistivity of a single SiCN nanowire is measured at room temperature for the first time. [ABSTRACT FROM AUTHOR]