
ALT Synthetic text, images, and video are rising. Software-only detectors are power-hungry, too slow, and costly, and do not scale with demand. When software cannot scale, historically hardware often comes to the rescue. We perform the first reverse engineering of a compression acceleratorβIntel QuickAssist Technology (QAT) microarchitecture andβfind different amount of time on each, a novel data structure-dependent property. Using this information, we then build Lasso, a low-cost, high-performance detector for AI-generated content that turns the accelerator into a measuring instrument. For text detection, Lasso achieves 100% accuracy on original DetectRL text and retains 92% average accuracy after paraphrasing, compared with 78% for Binoculars and 55% for DetectGPT. Across original, paraphrased, and perturbed text, it achieves 82% average accuracy, outperforming all five baselines. With topic, length, and formatting matched, accuracy reaches 95% for instruction-tuned generators, including

ALT Generating a token with a language model requires a fixed amount of arithmetic, whose energy sets a floor beneath any hardware. For Qwen3-1.7B at a 16K-token context, gate-level measurements at 5 nm place this floor at 0.89 mJ per token, yet an NVIDIA B300 spends 391-1,120x that. The excess is the price of generality, software layers, programmable hardware and moving weights and key-value (KV) caches from external memory, and it widens with each process generation. We present a co-design approach that brings token generation close to this floor and then lowers the floor itself, optimizing silicon and model together in a gate-level search loop. Fixing one model architecture in the datapath, with weights still loadable, and holding weights and every conversation's KV cache on-die, a pre-silicon 5 nm design brings an unchanged Qwen3-1.7B token to 3.4x its floor. Co-designing the model with the circuits at fixed quality lowers the floor to 0.29 mJ and the token to 0.52 mJ, 1.8x the new flo

ALT Wafer-scale accelerators offer massive compute capacity through hundreds of thousands of processing elements (PEs) interconnected across a single wafer, but realizing this potential requires efficient application mapping. Their scale makes programming these systems and mapping applications onto them challenging. Existing approaches rely on template-based mappings to maintain productivity, but these templates restrict the mapping design space. In particular, they impose fixed geometric layouts over the PE grid, often leaving compute and memory resources underutilized. In this work, we present WaferTetris, a system that removes these geometric restrictions through freeform mapping. Its Tetris-like strategy flexibly packs computation across the wafer, improving PE utilization and usable memory capacity while preserving PE-task code. We develop a recurrence-based throughput model that captures branch-and-join dependencies to guide spatial parallelism, and an architecture model with per-col

ALT Event-camera segmentation requires persistent neural state despite sparse input events, making memory organization a central constraint on embedded implementation. We present EventSpikeSeg, a mixed-signal architecture combining resident capacitor memory with shared analog processing in GF180MCU 180-nm CMOS. Its 25-population spiking encoder-decoder has 143,824 synaptic weights but requires 30,976,000 membrane states at 440 x 640 resolution. The mapping specifies chronological 5-ms model updates, digital event/address services and 64-state capacitor banks. Characterization covers 22 module families and 74 operating points, distinguishing transistor simulations from digital-library estimates. Nominal results include 41.711 microwatts biased neuron-core power, 3.338-pJ positive terminal energy for a 1.4-microsecond cell read, and 0.531-mV selected-state drift over a 5-ms extracted-tile hold. Commit and coupling errors are reported separately. The software checkpoint records 27.976% mean i

ALT Long-context LLMs must retain growing histories, but full attention makes each new token increasingly expensive. Inference-time learning (ITL) offers another approach. The model keeps training on the context during inference, retaining what it learns in its weights. Recent work shows that ITL can match full-context attention quality while attending only to a recent fixed-size window. This bounds attention costs but adds Learning (backward computation and weight updates) alongside Prediction (the forward pass) costs. Inference accelerators assume fixed weights, while training accelerators target complete training steps. This work identifies two opportunities. Attention's non-matrix operations leave enough idle matrix capacity for Learning, while selected layers can tolerate delayed weight updates with little quality loss. We exploit both with Learning While Serving (LWS), an inference accelerator architecture that shares compute hardware between latency-critical Prediction and deadline-

ALT Physical design automation of very large-scale integrated circuits (VLSI) has made significant strides, progressing from single-stage placement and routing toward increasingly integrated full-flow implementation. Yet, existing approaches struggle with the vast design space optimization (DSO) arising from diverse technology nodes, circuit designs, and varied constraints. The substantial runtime of physical-design tools further exacerbates this challenge, leaving industrial workflows heavily dependent on experienced engineers to navigate the design space and achieve power, performance, and area (PPA) targets, incurring significant engineering effort. We argue that large language models (LLMs) with strong reasoning capabilities, enriched with distilled human design expertise, offer a promising path toward efficient and automated DSO. Unlike traditional algorithms, LLMs can interpret complex design contexts and naturally incorporate human expertise expressed in natural language for reasoni

ALT Growing visual data overwhelms storage on resource-constrained edge devices, making compression essential. Traditional codecs offer inadequate rate-distortion performance, motivating the exploration of diffusion codecs. Diffusion codecs face two limitations: (1) At the algorithm level, diffusion decoders preserve perceptual quality under extreme compression, but require substantial computation. (2) At the architecture level, in NPU-centered devices, neural codecs occupy the NPU, interfering with other AI applications. Meanwhile, the complex codec pipeline incurs long I/O paths, leading to additional overhead and energy cost. To address these issues, we propose DiffusionSSD, a cross-level near-data processing framework for extreme visual data compression on edge devices, using an algorithm-architecture co-design method. Specifically, (1) at algorithm level, we present Dual-stage Diffusion, a lightweight solution tailored for diffusion codecs that employs patchification and single-step d

ALT Stack-based reconvergence enables efficient SIMT execution with a shared warp program counter and compact control-flow state. However, its fixed serialization of divergent paths can deadlock programs that rely on intra-warp synchronization: a spinning subgroup may prevent a resource-holding sibling from executing. Independent Thread Scheduling (ITS) enables finer-grained progress through per-thread execution state, but introduces additional hardware cost. We present SIMTurn, a lightweight compiler-hardware mechanism that enables forward progress among divergent subgroups while retaining stack-based reconvergence. SIMTurn combines a compiler-inserted yield instruction at potentially blocking loops with a bounded pool of resumable subgroups. The scheduler allows a waiting subgroup to yield to a sibling that can make progress, without requiring independent program counters for every thread. We implement SIMTurn in the compiler, simulator, and RTL on the open-source Vortex GPGPU and evalua

ALT Reconfigurable Dataflow Accelerators (RDAs) have demonstrated high performance and energy efficiency for dense, regular tensor applications. However, conventional RDAs with static configurations show dramatically reduced performance and efficiency on irregular tensor workloads with data-dependent, non-uniform, dynamic tensor shapes. We propose Elasticy, a dynamically reconfigurable dataflow accelerator that employs streaming and tile abstractions to minimize the overhead of dynamic support. By encoding dynamic shapes into streams using rank-variable expressions, Elasticy exploits data-dependent tensor shape dynamism to minimize off-chip data traffic and wasted compute cycles. Elasticy employs stream-embedded reconfiguration information to trigger fine-grained runtime scheduling decisions. Elasticy also introduces a novel distributed linked streaming memory (DLSM) that more effectively uses on-chip memory to minimize off-chip data traffic and support dynamically sized tensors. Both reco

ALT Edge AI increasingly relies on specialized accelerators for low-latency inference under tight resource and energy constraints, yet their efficiency remains limited by data movement. Prior work, CHARM, mitigates this overhead by overlapping data movement with computation. We observe that in-memory computing can further supplement tensor processing as a parallel accelerator while eliminating certain data movements entirely. We present CHARMING, a heterogeneous edge AI architecture that uses accuracy- and quantization-aware partitioning to distribute and concurrently execute inference workloads across tensor processing and in-memory computation. CHARMING partitions attention heads between an AIE array and an in-DRAM compute unit, balancing execution according to model architecture, sequence length, and precision while selecting heads to minimize accuracy degradation. CHARMING also integrates quantization into the AIE projection stage to further reduce data movement. We validate CHARMING e

ALT Large language model (LLM) inference at datacenter scale is fundamentally bottlenecked by memory capacity and bandwidth limits. Because model weights are read-only and identical across instances, conventional architectures that replicate these parameters across the local DRAM of tens of thousands of accelerators incur massive capital expenditures, significant static power dissipation, and redundant memory-fetch energy. A recent proposal to use optical fiber loops as delay-line memory alleviates these issues, but its recirculating loop must hold the whole model: loop length grows with model size, slack is fixed when the ring is built, and the layouts that serve modern Mixture of Experts (MoE) models efficiently do not fit in it. We propose Optical Broadcast Memory (OBM), an open-ended hierarchical Tap-and-Amplify fiber architecture that generalizes and replaces the optical ring paradigm. A centralized memory server continuously broadcasts weights to distributed inference nodes that pass

ALT Users of artificial intelligence services, including those powered by large language models and other transformer-based models, need assurance that inference and training follow the declared computation without exposing private models or data. Zero-knowledge proofs can provide this assurance, but proof generation remains expensive. Hardware acceleration research often reports gains for individual cryptographic operations, leaving their effect on the full cost of proving neural-network workloads, particularly large transformer models, unclear. We introduce a four-layer framework linking verifiable AI workloads, proof systems, cryptographic kernels and hardware platforms, use it to review the literature, and add original measurements of six provers and seven kernels to identify which costs matter in practice. Our measurements show that the dominant cost changes with the proof system and input length. In low-rank fine-tuning, the cost of opening commitments to fixed model weights remains

ALT SPEC CPU2026, the latest generation of the SPEC CPU benchmarks, is set to become the standard benchmark suite in computer architecture research, where new designs are often evaluated through sampled simulation with tools such as SimPoint. Existing characterizations of the suite focus on whole-run averages, which reveal neither how behavior changes within a run nor how well sampling captures it. We characterize CPU2026 on real hardware at three levels: programs, phases, and samples. At the program level, we report whole-run metrics and top-down bottleneck classes for all four suites and examine how they change with workload size, compiler version, and platform, across machines with three different instruction set architectures: x86, Arm, and RISC-V. Notably, changing only the compiler version can halve or nearly double the IPC of individual benchmarks. At the phase level, we track hardware counters every 10 ms over all 90 SPECrate inputs, identify recurring patterns of time-varying beha

ALT Symbolic reasoning at scale requires representations that are both expressive and hardware-efficient. Hyperdimensional Computing (HDC) provides a promising brain-inspired framework, encoding concepts as high-dimensional vectors composed through efficient algebraic operations. However, the widely used Fourier Holographic Reduced Representation (FHRR) relies on commutative binding, which cannot encode order or directionality. Prior FPGA/ASIC accelerators recover this through costly permutation logic, incurring irregular memory access and high control overhead. Generalized Holographic Reduced Representations (GHRR) instead redefine binding as non-commutative matrix multiplication, naturally encoding sequences, graphs, and hierarchies without permutations-but at the cost of turning every element into a matrix, introducing compute and storage overhead that limits scalability. In this work, we propose a flattening methodology that preserves GHRR's matrix-based encoding while executing traini

ALT Connected health devices increasingly support health monitoring outside clinical settings, making secure and accessible physiological data important for patients and caregivers. However, device-specific applications and interfaces complicate the integration of readings from different devices. The Matter standard provides standardized device commissioning and protected communication, but its cluster library does not represent the health attribute measurement. We present a preliminary demo study of a Matter-based IoMT testbed with a specific HealthVitals cluster. Implemented in the connectedhomeip SDK on a Raspberry Pi 5, the cluster exposes simulated heart rate, blood oxygen saturation, and body temperature. We examine commissioning, attribute retrieval, communication under simulated packet loss, and operational traffic through a reference controller. The study demonstrates that this custom telemetry model can operate within the tested Matter configuration without modifying commissionin

ALT Protocol bridge verification is a persistent bottleneck in SoC development: for every new bridge instance, engineers hand-craft a full UVM testbench from scratch, consuming 40β100 engineering hours of bring-up effort. We present an agentic Python framework that generates complete, protocol-aware UVM testbenches for arbitrary IP and subsystem configurations from a structured Verification Manifestβa JSON specification encoding interface topology, protocol semantics, and coverage intent. The framework employs a dual closed-loop architecture: an inner compiler feedback loop that injects categorised UVM diagnostics rather than raw tool output, and an outer coverage loop driven by intent-gap analysis against the Manifest. Validated on the AMBA AHB-to-APB bridgeβa representative production protocol bridge with a 7-state control FSM, two-phase APB handshake, and non-trivial burst serialisation logicβthe framework achieves 96.8% functional coverage, a 4.7Γ bring-up speedup, and a Protocol Verif

ALT Modern cloud platforms use memory bandwidth regulation to isolate co-located workloads, implicitly assuming that high aggregate bandwidth consumption implies high interference. We show that this assumption is fundamentally flawed. We present the Single-Bank Attack (SBA), a practical memory denial-of-service attack that induces extreme cross-core slowdownsβover three orders of magnitude on memory-intensive victims (e.g., LLM decode)βwhile consuming less than 1 GB/s of memory bandwidth. SBA exploits the gap between peak and worst-case DRAM throughput by concentrating memory write operations on a single bank, forcing persistent row conflicts and frequent write draining, thereby collapsing the memory service rate. Across Intel and AMD systems, including DDR5- and HBM2E-based platforms, SBA consistently causes orders-of-magnitude slowdowns under realistic multi-tenant settings, revealing a fundamental limitation of bandwidth-centric memory isolation. We further show that this localized cont

ALT The micro-Doppler (uD) spectrum provides detailed characteristics of target movement, making it highly valuable for applications such as target classification and human activity recognition. However, traditional Integrated Sensing And Communication (ISAC) systems are highly view-dependent and limited by blind spots and angle dependencies. Existing approaches rely on either configuration-specific solutions or data-driven models. The former results in poor generalization when device deployments or target orientations change, while the latter requires acquiring extensive multi-view datasets, which is time-consuming and not always feasible. To address these limitations, in this work we propose VISTA, a novel method for reconstructing the uD spectrogram in an absolute coordinate frame, yielding view-independent Doppler features that are invariant to the spatial distribution of ISAC nodes. We formulate the uD estimation as a sparse reconstruction problem that jointly leverages measurements f

ALT Modern DRAM chips are vulnerable to read disturbance, which manifests as bitflips in a victim row physically near an aggressor row that is repeatedly activated. State-of-the-art mitigations rely on accurate and exhaustive profiling of the read disturbance threshold (RDT), i.e., the minimum number of aggressor row activations needed to induce the first bitflip. However, RDT changes significantly and unpredictably over time, motivating online profilers that dynamically adjust a mitigation's RDT configuration. RDT profiling is costly for online profilers (e.g., hundreds of seconds for a 16-bank DDR4 module) because 1) measuring a single DRAM row's RDT requires many activations and 2) each DRAM row can have a different RDT, requiring many rows to be profiled. We introduce HARE, an online RDT profiling technique that intentionally exacerbates read disturbance to quickly profile a DRAM module. HARE's key idea is to reduce the timing constraints between consecutive DRAM commands during profil

ALT The increasing demand for real-time processing and personalized deep learning has made edge-device training essential but challenging due to high memory and computational costs. To address this, we propose a Convolutional Neural Network (CNN) training accelerator with a dynamic sparse update mechanism that selectively updates important parameters using a random channel strategy, avoiding costly evaluations. Combined with activation pruning, our approach reduces activation storage to just 3.4% of the original. Our hardware features a fused forward dataflow that concurrently processes group normalization and convolution, reducing training time by 45% through computation reordering and efficient zero-skipping. Implemented in TSMC 28nm at FP8 precision, the design operates at 800 MHz, achieving 8 TOPS peak throughput, 31.13 TOPS/W energy efficiency, and 14.8 TOPS/mmΒ² area efficiency, with power consumption ranging from 257 to 272 mW. The design enables on-device training within a sub-100KB
