Area efficiency is vital for floating-point units (FPUs) in resource-constrained IoT devices. However, existing designs suffer from rigid architectures and costly arithmetic units, limiting performance-area optimization. To this end, this work presents FPUltra, an area-efficient single-precision FPU for cost-sensitive RISC-V cores. FPUltra adopts a novel phase-decoupled control architecture to mitigate timing hazards and improve execution efficiency. A parallel approximate floating-point multiplier (FPM) is designed using combinational logic, based on the Mitchell algorithm with error compensation. A Newton–Raphson-based subinstruction decomposition method is presented to support floating-point division (Fdiv) and square root (Fsqrt). Compared with state-of-the-art FPUs, FPUltra achieves 9%–695% and 101%–14 186% improvements in equivalent slices efficiency (Eq.Slices Eff.) on FPGA and equivalent area efficiency (Eq.Area Eff.) on ASIC, respectively. Our code will be available athttps://github.com/LX-IC/FPUltra

Segment anything model (SAM), a foundation model with superior versatility and generalization across diverse segmentation tasks, has attracted widespread attention in medical imaging. However, it has been proved that SAM would encounter severe performance degradation due to the lack of medical knowledge in training and local feature encoding. Though several SAM-based models have been proposed for tuning SAM in medical imaging, they still suffer from insufficient feature extraction and highly rely on high-quality prompts. In this paper, we propose a powerful foundation model SAMCT allowing labor-free prompts and train it on a collected large CT dataset consisting of 1.1M CT images and 5M masks from public datasets. Specifically, based on SAM, SAMCT is further equipped with a U-shaped CNN image encoder, a cross-branch interaction module, and a task-indicator prompt encoder. The U-shaped CNN image encoder works in parallel with the ViT image encoder in SAM to supplement local features. Cross-branch interaction enhances the feature expression capability of the CNN image encoder and the ViT image encoder by exchanging global perception and local features from one to the other. The task-indicator prompt encoder is a plug-and-play component to effortlessly encode task-related indicators into prompt embeddings. In this way, SAMCT can work in an automatic manner in addition to the semi-automatic interactive strategy in SAM. Extensive experiments demonstrate the superiority of SAMCT against the state-of-the-art task-specific and SAM-based medical foundation models on various tasks. The code, data, and model checkpoints are available at https://github.com/xianlin7/SAMCT.

Design space exploration (DSE) is crucial for optimizing the performance, power, and area (PPA) of CPU microarchitectures ($\mu $-archs). While various machine learning (ML) algorithms have been applied to the$\mu $-arch DSE problem, the potential of reinforcement learning (RL) remains underexplored. In this article, we propose a novel RL-based approach to address the reduced instruction set computer V (RISC-V) CPU$\mu $-arch DSE problem. This approach enables dynamic selection and optimization of$\mu $-arch parameters without relying on predefined modification sequences, thus significantly enhancing exploration flexibility. To address the challenges posed by high-dimensional action spaces and sparse rewards, we use a discrete soft actor-critic (SAC) framework with entropy maximization to promote efficient exploration. In addition, we integrate multistep temporal-difference (TD) learning, an experience replay (ER) buffer, and return normalization to improve sample efficiency and learning stability during training. Our method further aligns optimization with user-defined preferences by normalizing PPA metrics relative to baseline designs. Experimental results on the Berkeley out-of-order machine (BOOM) demonstrate that the proposed approach achieves superior performance compared with state-of-the-art methods, showcasing its effectiveness and efficiency for$\mu $-arch DSE. Our code is available athttps://github.com/exhaust-create/SAC-DSE.

Despite the great potential in capturing long-range dependency, one rarely-explored underlying issue of transformer in medical image segmentation is attention collapse, making it often degenerate into a bypass module in CNN-Transformer hybrid architectures. This is due to the high computational complexity of vision transformers requiring extensive training data while well-annotated medical image data is relatively limited, resulting in poor convergence. In this paper, we propose a plug-n-play transformer block with dynamic token merging, named DTMFormer, to avoid building long-range dependency on redundant and duplicated tokens and thus pursue better convergence. Specifically, DTMFormer consists of an attention-guided token merging (ATM) module to adaptively cluster tokens into fewer semantic tokens based on feature and dependency similarity and a light token reconstruction module to fuse ordinary and semantic tokens. In this way, as self-attention in ATM is calculated based on fewer tokens, DTMFormer is of lower complexity and more friendly to converge. Extensive experiments on publicly-available datasets demonstrate the effectiveness of DTMFormer working as a plug-n-play module for simultaneous complexity reduction and performance improvement. We believe it will inspire future work on rethinking transformers in medical image segmentation. Code: https://github.com/iam-nacl/DTMFormer.

In the Internet of Things (IoT) field, cloud and fog computing dramatically increase the complexity of floating-point (FP) calculations. Cost-constrained microcontrollers (MCUs) urgently need more efficient FP computing methods, such as integrated FP units (FPUs). To this end, this brief proposes FPUx, a high-performance FPU designed through a hybrid pipeline and state-machine approach. The FPUx is integrated into E203 for implementation (E203-FPUx). Furthermore, the Easy-lite is proposed to reduce handshake delay and a range of single-precision FP (FP32) arithmetic IPs are designed to customize FPUs. Compared with E203-FPnew and E203, the performance of E203-FPUx is improved by$1.5\times $and$36\times $, and the total energy consumption is saved by 36% and 1430% on average, respectively.

OBJECTIVE: Transformers, born to remedy the inadequate receptive fields of CNNs, have drawn explosive attention recently. However, the daunting computational complexity of global representation learning, together with rigid window partitioning, hinders their deployment in medical image segmentation. This work aims to address the above two issues in transformers for better medical image segmentation. METHODS: We propose a boundary-aware lightweight transformer (BATFormer) that can build cross-scale global interaction with lower computational complexity and generate windows flexibly under the guidance of entropy. Specifically, to fully explore the benefits of transformers in long-range dependency establishment, a cross-scale global transformer (CGT) module is introduced to jointly utilize multiple small-scale feature maps for richer global features with lower computational complexity. Given the importance of shape modeling in medical image segmentation, a boundary-aware local transformer (BLT) module is constructed. Different from rigid window partitioning in vanilla transformers which would produce boundary distortion, BLT adopts an adaptive window partitioning scheme under the guidance of entropy for both computational complexity reduction and shape preservation. RESULTS: BATFormer achieves the best performance in Dice of 92.84 %, 91.97 %, 90.26 %, and 96.30 % for the average, right ventricle, myocardium, and left ventricle respectively on the ACDC dataset and the best performance in Dice, IoU, and ACC of 90.76 %, 84.64 %, and 96.76 % respectively on the ISIC 2018 dataset. More importantly, BATFormer requires the least amount of model parameters and the lowest computational complexity compared to the state-of-the-art approaches. CONCLUSION AND SIGNIFICANCE: Our results demonstrate the necessity of developing customized transformers for efficient and better medical image segmentation. We believe the design of BATFormer is inspiring and extendable to other applications/frameworks.

Vision transformers have recently set off a new wave in the field of medical image analysis due to their remarkable performance on various computer vision tasks. However, recent hybrid-/transformer-based approaches mainly focus on the benefits of transformers in capturing long-range dependency while ignoring the issues of their daunting computational complexity, high training costs, and redundant dependency. In this paper, we propose to employ adaptive pruning to transformers for medical image segmentation and propose a lightweight and effective hybrid network APFormer. To our best knowledge, this is the first work on transformer pruning for medical image analysis tasks. The key features of APFormer are self-regularized self-attention (SSA) to improve the convergence of dependency establishment, Gaussian-prior relative position embedding (GRPE) to foster the learning of position information, and adaptive pruning to eliminate redundant computations and perception information. Specifically, SSA and GRPE consider the well-converged dependency distribution and the Gaussian heatmap distribution separately as the prior knowledge of self-attention and position embedding to ease the training of transformers and lay a solid foundation for the following pruning operation. Then, adaptive transformer pruning, both query-wise and dependency-wise, is performed by adjusting the gate control parameters for both complexity reduction and performance improvement. Extensive experiments on two widely-used datasets demonstrate the prominent segmentation performance of APFormer against the state-of-the-art methods with much fewer parameters and lower GFLOPs. More importantly, we prove, through ablation studies, that adaptive pruning can work as a plug-n-play module for performance improvement on other hybrid-/transformer-based methods. Code is available at https://github.com/xianlin7/APFormer.