Convolution neural networks (CNNs) have been extensively used in machine learning applications. The most time-consuming part of CNNs are convolution operations. A common approach to implementing convolution operations is to recast them as general matrix multiplication, known as the im2col+GEMM approach. There are two main drawbacks of this approach. One is that large additional memory space is required. The other is the packing on the input elements of convolution operations are not memory-efficient enough. In this paper, we present a new parallel convolution algorithm using implicit matrix multiplication on multi-core CPUs. In comparison with Im2col+GEMM, our new algorithm can reduce the memory footprints and improve the packing efficiency. The experiment results on two ARV8-based multi-core CPUs demonstrate that our new algorithm gives much better performance and scalability than the im2col+GEMM method in most cases.

Libo Zhang, Zhaoning Zhang, Xubaizhou, Rui Li, Zhiliang Tian, Songzhu Mei, Dongsheng Li. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Cloud computing bring a tremendous complexity to information security. Remote attestation can be used to establish trust relationship in cloud. TBVMM is designed to extend the existing chain of trust into the software layers to support dynamic remote attestation for cloud computing. TBVMM uses Bayesian network and Kalman filter to solve the dynamicity of the trusted relationship. It is proposed to fill the trust gap between the infrastructure and upper software stacks.

At present, with the rapid development of big data processing technology, streaming data processing and real-time data analysis have gradually become new research hotspots. Both the industry and the academia have invested a lot of research into the efficient processing methods of massive data generated in the environment such as the Internet and e-commerce. Meanwhile, high-performance computing technology and supercomputers are also looking for new business growth points. The convergence of big data processing and high performance computing technology is the general trend of big data analysis in the future. This paper will give a brief overview of typical technologies in the fusion process of big data processing and high-performance computing.

The graph has recently enabled substantial advances in big data analysis. As graphs are increasing from billions to trillions, efficient graph processing requires large-scale distributed clusters, which have up to thousands of nodes. For big data applications of which the computation is relatively simple, while the communication, especially for imbalanced communication is the bottleneck on distributed clusters, where huge numbers of small messages are transferred through 2D-topology networks. Graph partitioning is the dominant factor to affect the performance of large-scale distributed graph processing. Current graph partitioning policies have paid extensive attention to the utilization of the power law of big graphs but failed to exploit the advanced architectural benefits of 2D topology. To address such a problem, this paper presents GraphMedia, a communication-balanced graph partitioning for distributed search at scale. The key idea of GraphMedia is a communication-balanced partitioning to balance communication based on hardware/software co-design, in which the power law of graphs would be explored to average communication among nodes, and communication would be balanced between row and column by leveraging advanced 2D-topology knowledge. We use both benchmarks and real-world graphs to validate GraphMedia. Specially, GraphMedia-based Graph500 tests on the Tianhe supercomputer are superior to the fastest systems in the latest Graph500 lists (June 2022). We finally apply GraphMedia to real-world graphs for online graph media access, which outperforms the state-of-the-art graph partitioning and graph system by orders of magnitude.

Cloud computing emerges as a change in the business paradigm that offers pay-as-you-go computing capability and brings enormous benefits, but there are numerous organizations showing hesitation for the adoption of cloud computing due to security concerns. Remote attestation has been proven to boost confidence in clouds to guarantee hosted cloud applications’ integrity. However, the state-of-the-art attestation schemes do not fit that multiple requesters raise their challenges simultaneously, thereby leading to larger performance overheads on the attester side. To address that, we propose an efficient and trustworthy concurrent attestation architecture under multi-requester scenarios, Astrape, to improve efficiency in the integrity and confidentiality protection aspects to generate an unforgeable and encrypted attestation report. Specifically, we propose two key techniques in this paper. The first one—aggregated attestation signature—reliably protects the attestation content from being compromised even in the presence of adversaries who have full control of the network, therefore successfully providing attestation integrity. The second one—delegation-based controlled report—introduces a third-party service to distribute the attestation report to requesters in order to save computation and communication overload on the attested party. The report is encrypted with an access policy by using attribute-based encryption and accessed by a limited number of qualified requesters, hence supporting attestation confidentiality. The experimental results show that Astrape can take no more than $0.4$ s to generate an unforgeable and encrypted report for 1000 requesters and deliver a throughput speedup of approximately $30\times$ in comparison to the existing attestation systems.