High-performance collective communication among GPUs in modern supercomputers is crucial for enabling many applications. Complex hierarchical interconnects between GPU devices necessitate collective algorithms that can effectively leverage the underlying network topology. We present parameterized algorithms for two GPU-to-GPU collectives, Allgather and Allreduce, as well as an optimized permutation kernel used to further enhance GPU collective communication. By employing a LogGP-based model calibrated with real machine measurements, we can efficiently simulate various parameter choices to identify optimal settings for specific device allocations and message sizes. Our comprehensive evaluation on NCSA Delta and Argonne Polaris supercomputers demonstrates that our parameterized algorithms can achieve, on average, a $20 \%$ speedup over their non-parameterized counterparts, with our parameter selection process capturing $98 \%$ of the potential speedup.

Exascale supercomputers have renewed the exigence of improving distributed communication, specifically MPI collectives. Previous works accelerated collectives for specific scenarios by changing the radix of the collective algorithms. However, these approaches fail to explore the interplay between modern hardware features, such as multi-port networks, and software features, such as message size. In this paper, we present a novel approach that uses system-agnostic, generalized (i.e., variableradix) algorithms to capture relevant features and provide broad speedups for upcoming exascale-class supercomputers.We identify hardware commonalities found on announced exascale systems and three omnipresent communication kernels (binomial tree, ring, and recursive doubling) that can be generalized to better leverage these features, creating 10 total implementations. For each kernel, we develop analytical models to intuit algorithm performance with varying radix values.Experiments on the world’s first exascale supercomputer (Frontier at ORNL) and a pre-exascale system (Polaris at ANL) show that our generalized algorithms outperform the baseline open-source and proprietary vendor MPI implementations by a significant margin, up to over 4.5x. We empirically determine optimal algorithms and parameter values, identifying where the analytical models are accurate and where hardware features directly determine performance. Most notably, we show how a single, system-agnostic implementation of a generalized algorithm can optimize for multiple hardware/software features across multiple systems.

The Domain Name System (DNS) is a key component of Internet-based communication and its privacy has been neglected for years. Recently, DNS over HTTPS has improved the situation by fixing the issue of in-path middleboxes. Further progress has been made with proxy-based solutions such as Oblivious DoH, which separate a user's identity from their DNS queries. However, these solutions rely on non-collusion between DNS resolvers and proxy networks. This paper instead proposes PDNS, a new DNS extension that uses Private Information Retrieval to allow DNS resolvers to operate on blind queries, thereby eliminating any privacy leaks.

In this paper, a distributed power system consisting of two power boards is designed for high-power distributed unmanned ground vehicle computer. This design scheme enables the power system to provide high-power supply at the same time to ensure a higher quality of power supply. Through the reasonable planning of power supply path and the accurate setting of decoupling capacitor, the power distribution network has higher voltage accuracy and lower voltage ripple noise. In addition, the Power Integrity simulation software is used to analyze, improve and verify the design to ensure that the power system design has a high quality. Finally, the actual measurement results show that the power system design scheme can effectively improve the power integrity of the distributed power network.

Motivated by improving the number of successful access for machine-to-machine (M2M) communications, and realizing resource allocation for any number of active M2M UEs with less computation, a computationally efficient adaptive resource allocation scheme is given in this paper. eNB first learns the resource allocation for few active M2M UEs, which is adaptively changed with the number of uplink resource blocks. Then the resource allocation for other M2M UEs is derived from those active M2M UEs. In addition, the access barring scheme is incorporated to improve the number of successful access. Experimental simulations demonstrate our scheme can significantly improve the number of successful access and reduce resource allocation computation.

This paper proposes a context-aware application infrastructure for the easy creation and flexible deployment of the context-aware application. In this infrastructure we introduce the concept of plane to manage the sensors and handle the environment information. Corresponding to the operations in this infrastructure we design a layered structure for the context information and define three kinds of working procedures. In addition, we apply the Demspter-Shafer evidence theory (DSET) to the reasoning mechanism of this infrastructure. Specially, based on combination rule of DSET, we construct the inferencer which is the core component for this infrastructure. In the end, we design an application scenario to demonstrate the operation of the inferencer.