1714 matching published research entries
Computer Science · 2026-10-30
arXiv:2610.38329v1 Announce Type: cross Abstract: Frontier large language models write correct CUDA kernels for most single-operator benchmark tasks, yet their accuracy collapses on accelerators whose architecture and programming language they have rarely seen. We present a case study of an LLM agent loop built for such accelerators.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.38262v1 Announce Type: cross Abstract: Fully Homomorphic Encryption (FHE) enables computation on encrypted data and is promising for privacy-preserving computing. However, FHE deployment remains constrained due to massive computational, memory, and data-movement overheads.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.38118v1 Announce Type: cross Abstract: Synapse-resolution fruit-fly (Drosophila) connectomes enable neuron-by-neuron simulation of entire insect nervous systems, yet whole-brain models still require computationally demanding software or multiple neuromorphic chips, and single FPGAs support only small or statistical networks.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.38110v1 Announce Type: cross Abstract: Deep neural networks (DNNs) are widely deployed in safety- and privacy-critical domains such as healthcare and finance, where models are trained on confidential data. This paper presents the first hardware-level attribute-stealing attack that bypasses state-of-the-art privacy defenses, revealing that accelerator optimizations intended for performance can directly…
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.40000v1 Announce Type: new Abstract: Smartphone agents increasingly work as teams, in which a large planner directs parallel workers across long sessions. Running such a team on the device requires 200B-class models and contexts that exceed phone DRAM many times over, yet existing tiers treat flash as storage: UFS starves per-token weight streaming, and mobile High-Bandwidth Flash (HBF) needs a redesigned…
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39882v1 Announce Type: new Abstract: Analog placement requires circuit-specific trade-offs among area, matching, routability, and electrical performance. Balancing these trade-offs through predefined objectives and fixed placement flows remains difficult.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39797v1 Announce Type: new Abstract: Modern processors rely heavily on data prefetching to bridge the growing performance gap between computation and memory accesses. In offset prefetching, prefetches are issued at an address delta relative to the most recently observed access, selecting deltas that have exhibited high coverage in the recent access-stream history.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39638v1 Announce Type: new Abstract: Signed digit redundant number systems such as Maximally Redundant Signed Digit (MRSD) enable carry-free addition but introduce significant overhead in multiplication due to the presence of both positive and negative bits. While Unsigned Digit Redundant (UDR) systems have shown promise in radix-2, their extension to higher radices remains unexplored.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39570v1 Announce Type: new Abstract: Long-run RTL simulation of CPU candidates can take hours, while short diagnostic runs may miss performance degradation that emerges later in execution. This evaluation cost can delay design decisions in LLM-agent-based CPU development.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39515v1 Announce Type: new Abstract: Analog layout automation remains challenging in Analog/Mixed-Signal circuit design, because parasitics and mismatches introduced during layout strongly affect circuit performance. Placement is especially critical among the layout stages, as it constrains the routing solution space and the achievable post-layout performance.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39420v1 Announce Type: new Abstract: Early screening of depression requires efficient analysis of audio and speech data, while conventional CPU and GPU platforms incur high latency and energy consumption for CNN-LSTM inference. To address this challenge, we propose a unified resistive RAM (ReRAM) processing-in-memory accelerator for a CNN-LSTM attention workload.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39395v1 Announce Type: new Abstract: Analog flows from specifications to a signed-off layout rarely run without manual iteration. Recent large language model (LLM) agents size circuits with optimizer loops over the device dimensions and stop at the schematic.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39380v1 Announce Type: new Abstract: AI infrastructure, the chips and software that train and serve AI models, is designed in layers. Layers make the system manageable, but they also fix which choices an optimizer can consider. Gains that need the chip and its software to change together are excluded before optimization begins: the formulation leaves them out.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39268v1 Announce Type: new Abstract: The computational cost of conventional Convolutional Neural Networks (CNNs) limits their deployment in low-power devices for high-speed edge AI applications. This paper presents an end-to-end cosine-domain CNN (CD-CNN) architecture for energy-efficient FPGA inference using high-level synthesis (HLS).
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.39040v1 Announce Type: new Abstract: The efficiency of Deep Neural Network (DNN) accelerators depends on how each tensor operation is mapped onto the hardware. A mapping specifies how the computation is tiled across the memory hierarchy, ordered, and parallelized across the processing elements. These choices form a combinatorially large search space, where efficient mappings are hard to find.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.38575v1 Announce Type: new Abstract: Achieving timing closure in FPGA designs is crucial, especially for high-performance applications. The Vivado Design Suite offers several strategies for optimization, but finding the best configuration for a given project remains challenging.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.38489v1 Announce Type: new Abstract: Register-Transfer Level (RTL) simulators are fundamental to modern IC design workflows, yet like any complex software, they frequently contain critical bugs that compromise simulation accuracy.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.38426v1 Announce Type: new Abstract: Analog layout remains largely manual and hard to automate. Generator frameworks such as gLayout express layouts as programs, producing layouts from parameterized, design-rule-aware generators; however, writing such generators still takes expert effort.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.37710v1 Announce Type: new Abstract: SumCheck is a central prover primitive in zero-knowledge proofs, exposing table-parallel work that maps naturally to GPUs but becomes increasingly arithmetic-intensive as finite fields widen. We study how this shift affects the cost of programming SumCheck in Triton rather than CUDA.
arXiv Computer Science ARRead article →Computer Science · 2026-10-30
arXiv:2610.37706v1 Announce Type: new Abstract: Recent advances in large language models (LLMs) have improved Verilog generation. However, most training pipelines distill natural-language reasoning traces whose intermediate design decisions are weakly structured and are primarily selected according to the correctness of the final generated RTL.
arXiv Computer Science ARRead article →Science · 2026-10-30
Environmental pollutants containing chlorine or bromine have been widely studied, but compounds containing both elements have remained largely unexplored. Researchers have now detected bromochloropyrene in industrial soils and harbor sediments, investigated its breakdown under simulated sunlight, and measured its activity through a biological receptor linked to toxic effects.
Phys.org ResearchRead article →Science · 2026-10-30
Global change is also affecting the behavior and mobility of the ocellated lizard, a reptile typical of Mediterranean ecosystems. This species—the largest lizard on the Iberian Peninsula—could experience changes in its ability to move efficiently between shelters and feeding areas, escape predators and interact with its natural environment.
Phys.org ResearchRead article →Science · 2026-10-30
Fish schools are a marvel of navigation. They move in concert without any one fish leading the way. Instead, their coordinated movements arise from individual fish continually adjusting their movements based on their neighbors. But how much of this social information each fish processes at once is less well understood.
Phys.org ResearchRead article →Science · 2026-10-30
Cells are very particular about the proper locations of their contents. For example, DNA is packaged in the nucleus in a highly specific way, but researchers are still trying to understand how this happens. For stem cells in the brain of a larval fruit fly—a model organism used to study processes such as early development—the DNA is organized so that the center of each strand, called the centromere, is at the top of…
Phys.org ResearchRead article →