Deep learning demonstrates effectiveness across a wide range of tasks. However, the dense and over-parameterized nature of these models results in significant resource consumption during deployment. In response to this issue, weight pruning, particularly through$N: M$sparsity matrix multiplication, offers an efficient solution by transforming dense operations into semisparse ones.$N: M$sparsity provides an option for balancing performance and model accuracy, but introduces more complex programming and optimization challenges. To address these issues, we design a systematic top-down performance analysis model for$N: M$sparsity. Meanwhile, NM-SpMM is proposed as an efficient general$N: M$sparsity implementation. Based on our performance analysis, NM-SpMM employs a hierarchical blocking mechanism as a general optimization to enhance data locality, while memory access optimization and pipeline design are introduced as sparsity-aware optimization, allowing it to achieve close-to-theoretical peak performance across different sparsity levels. Experimental results show that NM-SpMM is 2.1x faster than nmSPARSE (the state-of-the-art for general$N: M$sparsity) and 1.4× to 6.3× faster than cuBLAS's dense GEMM operations, closely approaching the theoretical maximum speedup resulting from the reduction in computation due to sparsity. NM-SpMM is open source and publicly available at https://github.com/M-H482/NM-SpMM.

Dynamic-shape tensor computation poses challenges for shape-specific compilation due to variable input dimensions. Existing compilers rely on shape samples, incurring high tuning costs and performance degradation on unseen inputs. We present Helix, a dynamic tensor compilation framework with sample-free compilation and architecture-guided optimization to achieve both compilation efficiency and shape-general performance. To avoid shape sampling, Helix constructs shape-agnostic compilation by decomposing computations across architectural layers. A bidirectional strategy combines top-down abstraction to align tensor computations with architectural hierarchies, and bottom-up kernel construction to build efficient execution strategies from reusable, architecture-aligned micro-kernels. A hybrid analyzer ensures accuracy through profiling at lower architectural levels, and achieves scalability through architecture-informed modeling at higher levels and runtime. This hierarchical design eliminates shape-specific tuning and enables shape-adaptive execution. Evaluations conducted on x86 CPUs, ARM CPUs, and NVIDIA GPUs demonstrate that Helix reduces compilation time by 174 × over the existing compilers and delivers 2.26 × and 3.29 × execution speedups over vendor libraries and dynamic-shape compilers, respectively.

This paper presents an open-source library that pushes the limits of performance portability for irregular General Matrix Multiplication (GEMM) on the widely-used Arm architectures. Our library, autoGEMM, is designed to support a wide range of Arm processors: from edge devices to HPCgrade CPUs. autoGEMM generates optimized kernels for various hardware configurations by auto-combining fragments of autogenerated micro-kernels that employ hand-written optimizations to maximize computational efficiency. We optimize the kernel pipeline by tuning the register reuse and the data load/store overlapping. In addition, we use a dynamic tiling scheme to generate balanced tile shapes. Finally, we position autoGEMM on top of the TVM framework where our dynamic tiling scheme prunes the search space for TVM to identify the optimal combination of parameters for code optimization. Evaluations on five different classes of Arm chips demonstrate the advantages of autoGEMM. For small matrices, autoGEMM achieves 98% of peak and up to 2.0x speedup over state-of-the-art libraries such as LIBXSMM and LibShalom. For irregular matrices (i.e. tall skinny and long rectangles), autoGEMM is 1.3-2.0x faster than widely-used libraries such as OpenBLAS and Eigen. autoGEMM is publicly available at: https://github.com/wudu98/autoGEMM.

Microservice architecture is a design philosophy that achieves decoupling by decomposing a monolithic application into multiple lightweight microservices. Meanwhile, edge computing can significantly reduce service latency and network congestion by extending computation and storage resources to the network edge. Therefore, in the microservice-oriented edge computing platform, a fundamental problem is how to efficiently deploy microservices with complex dependencies on the resource-constrained edge servers to satisfy the Quality of Service (QoS) constraints of users. Most of the existing studies ignore multiple call graphs with differentiated dependencies for an application, which often result in the violation of QoS. To address this issue, in this article, we first model the request response time of multiple instances and multiple call graphs scenario with service conflicts. Then, different from the existing heuristic or approximation algorithms which rely heavily on expert knowledge, we propose a graph-reinforcement-learning-based deployment (GRLD) framework. GRLD uses a graph convolutional network (GCN) to extract the graph data required for multiple call graphs with messages passing and aggregation, and the generated feature is fed into the underlying network of deep-reinforcement-learning (DRL). Experimental results show that GRLD outperforms counterparts in reducing service deployment overhead while satisfying QoS constraints of multiple call graphs.

The advancement of Offline Reinforcement Learning (RL) and Offline Multi-Agent Reinforcement Learning (MARL) critically depends on the availability of high-quality, pre-collected offline datasets that represent real-world complexities and practical applications. However, existing datasets often fall short in their simplicity and lack of realism. To address this gap, we propose Hokoff, a comprehensive set of pre-collected datasets that covers both offline RL and offline MARL, accompanied by a robust framework, to facilitate further research. This data is derived from Honor of Kings, a recognized Multiplayer Online Battle Arena (MOBA) game known for its intricate nature, closely resembling real-life situations. Utilizing this framework, we benchmark a variety of offline RL and offline MARL algorithms. We also introduce a novel baseline algorithm tailored for the inherent hierarchical action space of the game. We reveal the incompetency of current offline RL approaches in handling task complexity, generalization and multi-task learning.

By deploying computing, storage, and bandwidth resources at the user side, vehicular edge computing (VEC) provides low-delay services for vehicle users. However, due to the limited resources of edge servers, how to efficiently meet the Quality-of-Service (QoS) requirements of multiple tasks and save the total energy consumption in a dynamic environment is an important issue in VEC. In this article, we first propose an energy consumption and QoS-aware co-offloading model. Unlike most previous studies, our goal is to minimize the total energy consumption while guaranteeing the QoS constraints of tasks, thus avoiding the overallocation of resources and high energy consumption caused by the one-sided pursuit of delay minimization. Then, without the requirements for domain experts, we propose Bayesian optimization-based computation offloading (BOCO) method to find the optimal offloading decision. To the best of our knowledge, this work is the first to apply Bayesian optimization to computation offloading in VEC. Furthermore, we conduct a series of experiments and comparisons with other offloading methods to analyze the effectiveness and performance of the proposed algorithm. Experimental results verify that our proposed BOCO outperforms counterparts.

In this paper, we propose a highly efficient computing method for game character control with phase-functioned neural networks (PFNN). The primary challenge to accelerate PFNN on mobile platforms is that PFNN dynamically produces weight matrices with an argument, phase, which is individual to each game character. Therefore existing libraries that generally assume frozen weight matrices are inefficient to accelerate PFNN. The situation becomes even worse when multiple characters are present. To address the challenges, we reformulate the equations and leverage the deep learning compiler stack TVM to build a cross-platform, high-performance implementation. Evaluations reveal that our solutions deliver close-to-peak performance on various platforms, from high-performance servers to energy-efficient mobile platforms. This work is publicly available at https://github.com/turbo0628/pfnn_tvm.

This paper introduces Honor of Kings Arena, a reinforcement learning (RL) environment based on the Honor of Kings, one of the world’s most popular games at present. Compared to other environments studied in most previous work, ours presents new generalization challenges for competitive reinforcement learning. It is a multi-agent problem with one agent competing against its opponent; and it requires the generalization ability as it has diverse targets to control and diverse opponents to compete with. We describe the observation, action, and reward specifications for the Honor of Kings domain and provide an open-source Python-based interface for communicating with the game engine. We provide twenty target heroes with a variety of tasks in Honor of Kings Arena and present initial baseline results for RL-based methods with feasible computing resources. Finally, we showcase the generalization challenges imposed by Honor of Kings Arena and possible remedies to the challenges. All of the software, including the environment-class, are publicly available.

We presentFastConv, a template-based code auto-generation open-source library that can automatically generate high-performance deep learning convolution kernels of arbitrary matrices/tensors shapes. FastConv is based on the Winograd algorithm, which is reportedly the highest performing algorithm for the time-consuming layers of convolutional neural networks. ARM CPUs cover a wide range of designs and specifications, from embedded devices to HPC-grade CPUs. The leads to the dilemma of how to consistently optimize Winograd-based convolution solvers for convolution layers of different shapes. FastConv addresses this problem by using templates to auto-generate multiple shapes of tuned kernels variants suitable for skinny tall matrices. As a performance portable library, FastConv transparently searches for the best combination of kernel shapes, cache tiles, scheduling of loop orders, packing strategies, access patterns, and online/offline computations. Auto-tuning is used to search the parameter configuration space for the best performance for a given target architecture and problem size. Results show 1.02x to 1.40x, 1.14x to 2.17x, and 1.22x and 2.48x speedup is achieved over NNPACK, ARM NN, and FeatherCNN on Kunpeng 920. Furthermore, performance portability experiments with various convolution shapes show that FastConv achieves 1.2x to 1.7x speedup and 2x to 22x speedup over NNPACK and ARM NN inference engine using Winograd on Kunpeng 920. CPU performance portability evaluation on VGG–16 show an average speedup over NNPACK of 1.42x, 1.21x, 1.26x, 1.37x, 2.26x, and 11.02x on Kunpeng 920, Snapdragon 835, 855, 888, Apple M1, and AWS Graviton2, respectively.

Deep Learning is ubiquitous in a wide field of applications ranging from research to industry. In comparison to timeconsuming iterative training of convolutional neural networks (CNNs), inference is a relatively lightweight operation making it amenable to execution on mobile devices. Nevertheless, lower latency and higher computation efficiency are crucial to allow for complex models and prolonged battery life. Addressing the aforementioned challenges, we propose FeatherCNN- a fast inference library for ARM CPUs - targeting the performance ceiling of mobile devices. FeatherCNN employs three key techniques: 1) A highly efficient TensorGEMM (generalized matrix multiplication) routine is applied to accelerate Winograd convolution on ARM CPUs, 2) General layer optimization based on custom high performance kernels improves both the computational efficiency and locality of memory access patterns for non-Winograd layers. 3) The framework design emphasizes joint layer-wise optimization using layer fusion to remove redundant calculations and memory movements. Performance evaluation reveals that FeatherCNN significantly outperforms state-ofthe-art libraries. A forward propagation pass of VGG-16 on a 64-core ARM server is 48, 14, and 12 times faster than Caffe using OpenBLAS, Caffe2 using Eigen, and NNPACK, respectively. In addition, FeatherCNN is 3.19 times faster than the recently released TensorFlow Lite library on an iPhone 7 plus. In terms of GEMM performance, FeatherCNN achieves 14.8 and 39.0 percent higher performance than Apple's Accelerate framework on an iPhone 7 plus and Eigen on a Samsung Galaxy S8, respectively. The source code of FeatherCNN library is publicly available at https://github.com/tencent/feathercnn.