Modern processors are equipped with single instruction multiple data (SIMD) instructions for fine-grained data parallelism. Compiler auto-vectorization techniques that target SIMD instructions face performance limitations due to insufficient information available at compile time, requiring programmers to manually manipulate SIMD instructions. SIMD intrinsics, a type of built-in function provided by modern compilers, enable programmers to manipulate SIMD instructions within high-level programming languages. Bugs in compilers for SIMD intrinsics can introduce potential threats to software security, producing unintended calculation results, data loss, program crashes, etc.

Large Language Models for code (code LLMs) have witnessed tremendous progress in recent years. With the rapid development of code LLMs, many popular evaluation benchmarks, such as HumanEval, DS-1000, and MBPP, have emerged to measure the performance of code LLMs with a particular focus on code generation tasks. However, they are insufficient to cover the full range of expected capabilities of code LLMs, which span beyond code generation to answering diverse coding-related questions. To fill this gap, we propose InfiBench, the first large-scale freeform question-answering (QA) benchmark for code to our knowledge, comprising 234 carefully selected high-quality Stack Overflow questions that span across 15 programming languages. InfiBench uses four types of model-free automatic metrics to evaluate response correctness where domain experts carefully concretize the criterion for each question. We conduct a systematic evaluation for over 100 latest code LLMs on InfiBench, leading to a series of novel and insightful findings. Our detailed analyses showcase potential directions for further advancement of code LLMs. InfiBench is fully open source at https://infi-coder.github.io/infibench and continuously expanding to foster more scientific and systematic practices for code LLM evaluation.

In view of the qualitative digitization and quantitative analysis of the damage status of ancient relics, this paper presents a method to automatically extract the surface damage area of the Zhengding ancient city wall. The point clouds of the damaged area were detected from the acquired laser point clouds using spatial change detection. The micro-deformation damage area is supplemented by extracting the local feature of the point cloud model. The experimental results show that the damaged region can be extracted effectively based on change detection and geometrical features, and the quantification and distribution of the damage status of cultural relics can be comprehensively expressed, which can provide important data for the renovation and restoration of relics.

This paper investigates the optimization of the generalized likelihood ratio test (GLRT) eigenvalue-based spectrum sensing detector in terms of decision thresholds and sensing time. In order to guarantee the interests of primary and secondary users simultaneously, the sensing performance is assessed using the total error rate, i.e., the summation of probabilities of false alarm and missed detection. Therefore, the generalized statistical distributions of the test statistic are derived under the absence and presence of primary users, assuming an arbitrary number of receive antennas. These distributions are necessary for the analyses of the total error rate performance and the optimization. The optimization consists of two parts. Firstly, the optimal decision threshold is numerically obtained, which can minimize the total error rate under the constraints of target probabilities of false alarm and missed detection. Secondly, the optimal sensing time is obtained when a target total error rate is guaranteed, so that the spectrum sensing process can be accelerated without the loss of sensing accuracy. Furthermore, the simulation and theoretical results reveal that the chosen optimal decision thresholds benefit the primary and secondary users simultaneously and the chosen optimal sensing time improves the speed of spectrum sensing.

This paper investigates optimization of the sensing threshold that minimizes the total error rate (i.e., the sum of the probabilities of false alarm and missed detection) of eigenvalue-based spectrum sensing techniques for multiple-antenna cognitive radio networks. Four techniques are investigated, which are maximum eigenvalue detection (MED), maximum minimum eigenvalue (MME) detection, energy with minimum eigenvalue (EME) detection, and the generalized likelihood ratio test (GLRT) detection. The contribution of this paper is of four parts. Firstly, we present the derivative of the matrix-variate confluent hypergeometric function, which is required for the MED case. Secondly, we derive the probabilities of false alarm for both cases MME and EME detection. Thirdly, we derive the probability of missed detection for the GLRT detector. Finally, we provide the exact expressions required to obtain the optimal sensing thresholds for all cases. The simulation results reveal that for all the investigated cases the chosen optimal sensing thresholds achieve the minimum total error rate.