In this work, we propose a view on electronic Neuromorphic Design Automation (eNDA), which we see as a design automation flow that bridges computational neuroscience modeling with traditional Electronic Design Automation (EDA) flow. We introduce the term, give examples of how it can be implemented, and design a prototype implementation: Syn2Logic. Syn2Logic is an entire eNDA framework, that allows neuroscientists to model neural behavior using a custom DSL and a compiler that takes the same model description down to synthesizable RTL hardware. We end the paper by applying the eNDA-flow through Syn2Logic to show how to -- without writing a single line of hardware description language (HDL) code-- (i) generate what we believe is the fastest C. elegans accelerator that runs significantly faster than state-of-the-art simulators, (ii) create (to the best of our knowledge) the fastest, most generic neuromorphic sudoku solver that outperforms CP-SAT and SCIP on TOP1465 puzzles, and (iii) create a 5.6 million FPS/Watt accelerator on a tiny FPGA that outperforms existing neuromorphic architectures in terms of speed and energy-efficiency on the MNIST dataset.
We ask what gets a language model onto the Apple Neural Engine (ANE) and what makes it fast there, and we answer with three measurements. We sweep a 64-shape matrix of LLM primitives that varies how a computation is expressed while holding what it computes fixed, recording per-operation device support. We then train matched models across size and precision, with quantized checkpoints byte-identical in structure to their fp16 counterparts, so every deployment measurement is of a real trained artifact. And we read the ANE's memory-controller byte counters during inference, establishing what actually ran rather than what the compiler intended. We support every headline claim with at least two of these three measurement paths. We find that placement is a property of how a computation is expressed, not of what it computes: a fused RMSNorm is fully ANE-eligible while its arithmetically identical decomposition is CPU-only. Weight encoding gates the accelerator: CoreML assigns a 25.85M-parameter conv-heavy fp16 model entirely to the CPU (our counters confirm zero bytes through the engine), while the same graph in int8 or 2-bit returns to ~83% residency and runs 1.8-2.2x faster, and a smaller 22.29M all-attention fp16 model sits at 98.9%. Decode cost is bytes streamed per token, at a constant ~0.77 fraction of nominal encoding width across fp16, int8 and 2-bit. The smallest and fastest models we measured are ternary, and at matched size the operator mix barely moves either axis: every resident 25M ternary model lands within 10.0-10.8 MB and 0.62-0.64 ms/token. The headline pair is half-attention ternary at 25M (10.5 MB, 0.63 ms) and 50M (16.8 MB, 0.86 ms) - 9.8x and 6.1x smaller, 3.0x and 2.2x faster than the conv-heavy fp16 design this work began with. From these measurements we draw a design procedure: choose the encoding first, then spend the byte budget on parameters.
Jiahao Lin, Alish Kanani, Sangwan Lee +2cs.AR cs.AI cs.LG
Hybrid Transformer-Mamba large language models (LLMs) enhance long-context efficiency, but their heterogeneous computation and communication patterns complicate efficient hardware acceleration. Chiplet-based architectures offer a scalable solution by integrating specialized compute and memory units. However, the design space spanning static architectural configurations and dynamic runtime policies is prohibitively large to explore exhaustively. To address this challenge, we present HYDRA, a comprehensive design space exploration framework for hybrid LLM serving on heterogeneous chiplet systems. HYDRA jointly explores chiplet composition, placement, inter-chiplet bandwidth provisioning, dynamic batching, and runtime scheduling. It integrates communication-aware placement, dynamic batching, elastic task scheduling, and a fast Markov-based performance estimator that captures multi-tenant runtime dynamics for efficient and accurate exploration. Across all workloads, HYDRA delivers 1.55x the throughput and 43.7 percent lower time-to-first-token on average, with throughput gains reaching up to 2.3x compared to state-of-the-art baselines. These results highlight that co-designing architecture and runtime policies is critical for efficient large-scale LLM serving on heterogeneous chiplet systems.
Reyhaneh Hosseinzadeh, Parham Zilouchian Moghaddam, Mehdi Modarressics.CV cs.AI cs.AR
The emergence of transformer-based deep learning models has brought unprecedented performance across various domains, particularly in natural language processing and computer vision. However, deploying these models, especially on resource-constrained devices, poses significant challenges due to their high computational complexity and large memory size and bandwidth requirements. This complexity has led researchers to use low-bit model weights to reduce memory usage and improve efficiency. In addition to reducing processing and memory demands, quantization introduces another useful property: value locality, where the extremely large number of parameters are restricted to a limited range of values. To fully take advantage of this locality, this paper presents DeVIT, an acceleration method for vision transformers that leverages differential computation to enable multiplier-less matrix multiplication.
Oliver Cassidy, Marta Andronic, George A. Constantinidescs.AR cs.LG
Mapping neural networks to FPGAs enables low-latency, energy-efficient inference, particularly for lookup table (LUT)-based models that eliminate multipliers and map directly to reconfigurable fabric. While prior work achieves high compute efficiency, it typically assumes full-sample availability, causing pipeline stalls in bandwidth-limited streaming scenarios. Here, the bottleneck shifts from computation to data movement, as large input transfers limit throughput and energy efficiency. We present CascadeLUT, an information-structured inference framework organized around bandwidth constraints. Instead of buffering the full input, features are partitioned into ordered subsets and predictions are progressively refined as subsets arrive. The cascade statically controls which layers consume incoming features, enabling deterministic streaming inference without runtime branching. By co-designing feature scheduling with hardware dataflow, CascadeLUT reduces data movement while maintaining accuracy. Across datasets, it achieves 4.0 to 12.5 times lower latency, 3.0 to 5.0 times higher throughput and up to 13.8 times lower energy/sample than prior LUT baselines, using 1.2 to 4.4 times the LUTs of the smallest DWN baseline per task. We also demonstrate on-device input quantization integrated with LUT-based inference and present end-to-end FPGA results on real-world workloads, with 5 times reductions in quantization overhead.
Fine-grained weight pruning and activation sparsification have emerged as effective approaches for reducing the compute and memory cost of inference for Transformer models. In the moderate-sparsity regime, Gustavson's dataflow provides a natural execution model for exploiting both activation and weight sparsity on vector processors through metadata-driven indexed accumulation. However, existing RVV architectures lack native support for this pattern, forcing kernels to rely on software index decoding and L1-backed indexed memory operations that keep sparse tensor contractions far below their roofline performance bound. We present Ventaglio, a runtime-configurable sparse execution unit coupled with RVV ISA extensions that drives sparse tensor contractions toward their roofline through indexed gather-accumulate-scatter support. Integrated into an open-source vector processing cluster and implemented in 12nm FinFET, Ventaglio accelerates sparse tensor contraction kernels by $6.9\text{--}7.4\times$ over optimized RVV baselines, with only $3.1\%$ area overhead for a cluster of tightly-L1 coupled vector processing elements. We build a performance-accurate instruction-level model of the Ventaglio extension, calibrate it against RTL implementation, and leverage it for scale-out performance analysis on a large $4\times4$ multi-cluster system. Using a DuoGPT-pruned LLaMA-3-8B model with practical $40\text{--}60\%$ dual sparsity, Ventaglio achieves $2.40\text{--}5.25\times$ and $2.06\text{--}3.16\times$ speedup over dense baselines during prefill and autoregressive decoding, respectively.
With the rapid adoption of long-context large language models (LLMs), the continuously growing KV cache during decoding has become the critical memory bottleneck. To tackle this challenge, we propose HiKV, a novel algorithm-hardware co-design that exploits KV cache redundancy through hierarchical importance awareness. Algorithmically, HiKV compresses the KV cache at two granularities: Stage I evicts unimportant tokens within a fixed budget, and Stage II further loads only the significant elements of each retained token, reaching compression ratios unattainable at a single granularity. Architecturally, we develop a dedicated accelerator centered on a reconfigurable importance sorter that switches between the distinct sorting datapaths each stage requires, unifying the two-stage acceleration in one circuit with minimal overhead. Evaluated on representative LLMs, HiKV achieves up to 7.95x speedup and 90% energy reduction in the attention computation over the vanilla KV cache baseline within negligible 1% accuracy loss. Under iso-accuracy constraints, HiKV outperforms state-of-the-art importance-based methods by achieving an additional 1.82~4.87x reduction in external memory accesses. These benefits are enabled by specialized hardware components that add only 8% to the system area.
Multi-task inference models share a single backbone across diverse tasks, yet execute identical computation regardless of which task is active - wasting energy and cycles on task-irrelevant operations. We observe that the task command, typically available before inference begins, provides a free signal that can be exploited to skip unnecessary computation at the hardware level. We present a HW/SW co-designed approach in which a lightweight gating network, trained jointly with the backbone, predicts per-tile binary execution masks conditioned on the task input. Each tile corresponds to a fixed group of output channels (the native scheduling granularity of the accelerator), enabling masked tiles to be skipped with zero overhead. This yields a task-dependent reduction in compute, where each command activates only the subset of the network it requires, without changes to the model architecture or inference pipeline. We co-design the full system stack: a command-conditioned training procedure that learns hardware-aligned tile masks under a sparsity objective; an instruction set architecture whose instructions carry per-tile bitmask fields, allowing the hardware to skip masked tiles without software intervention; and a tiled inference accelerator with configurable parallelism, double-buffered memory, and INT8 datapath that natively supports sparse tile execution. We prototype on an AMD/Xilinx Alveo U50 FPGA and evaluate on a closed-loop visuomotor driving task in CARLA autonomous driving simulator. Task-conditional sparsity reduces FLOPs by 66-76% while maintaining driving quality. On-device latency decreases by 51-59%, from 9.12 ms to 3.74-4.44 ms (2.1-2.4x speedup), with energy per inference dropping from 263 to 108-128mJ.
Transformers are widely used across many domains, including natural language processing, computer vision, web search, and DNA sequence analysis. Given their broad applicability, improving the performance of transformer models is critical. However, the high volume of data movement between processing units and memory during attention operations significantly limits their efficiency. Processing-In-Memory (PIM) mitigates this issue by performing computations directly inside memory. While prior work has proposed PIM-based transformer implementations, they suffer from costly inter-bank communication, and struggle to scale due to the limited capacity of memory banks. As a result, attention-related data must be split across banks, diminishing the potential benefits of PIM. In this work, we propose RED-PIM, an algorithm-architecture co-design that reduces attention latency by minimizing inter-bank data movement from O(N^2) to O(N) and shrinking intermediate attention matrices from N x N to d x d. By reorganizing matrix operations, performing computations locally, and employing an optimized data transfer strategy, RED-PIM significantly reduces computation cost and interconnect traffic. Compared to baseline PIM implementation, RED-PIM achieves inference time reductions ranging from 16.05% to 99.99% (geometric mean of 66.42%), with the largest gains on longer sequences. On real-world datasets, RED-PIM improves performance by 99.60% for long documents and 13.44% for shorter ones, while maintaining or improving accuracy. These results demonstrate RED-PIM's effectiveness for scalable and efficient transformer inference.
Nabila Tasnim, Haoran Liu, Qing Cao +1cs.AR cs.ET cs.LG
Several edge computing platforms, such as autonomous vehicles and smart sensing devices, need to adapt to dynamic environments in real time by learning from new data in the field. Continual learning has emerged as a promising solution for edge training, by incorporating techniques that successfully combine a highly summarized version of previously trained data (to avoid catastrophic forgetting) with recently sensed data. However, as is the case with other ML algorithms, continual learning generates significant data movement between general-purpose CPUs/GPUs and memory, impacting the suitability of continual learning for edge platforms. In-memory computing (IMC; also known as processing-using-memory) can curtail this waste and make continual learning feasible at the edge, but it faces two unique challenges: (1) IMC architectures make use of noisy computation operations that significantly harm training accuracy; and (2) IMC architectures have poor and often incomplete support for resource-efficient training. To address these challenges, we propose CLASP (the Continual Learning Acceleration System Platform), which to our knowledge is the first end-to-end system with IMC acceleration for continual learning. The hardware and software of CLASP are co-designed to support a wide range of continual learning algorithms, through software-visible assembly-level instructions that can be incorporated without constraints into ML-based algorithms. CLASP is designed around a back-end-of-line (BEOL) compatible ECRAM device that we fabricate, which can overcome the challenges of IMC-based training using other emerging memory devices. We show that CLASP with ECRAM approaches the accuracy of in-GPU training, while delivering a speedup of 67x and energy savings of 132x for learning without forgetting and experience replay using MNIST.
Nicholas Fry, Ignacio Alzugaray, Mark Pupilli +2cs.GR cs.CV cs.DC
We present the first implementation of a 3D Gaussian renderer on an Intelligence Processing Unit (IPU), comprising 1,472 independent tiles with only on-chip SRAM; constraints that approximate properties of efficient sensor-processor architectures. Our input scenes are 3D Gaussian maps from real-world sequences. Each tile 'owns' a screen-space region of the framebuffer; Gaussian primitives are routed to destination tiles via Manhattan-distance hops on a north-east-west-south (NEWS) grid, then distributed to overlapping neighbours in an expanding tree pattern. Computation follows the IPU's Bulk Synchronous Parallel (BSP) model, with inter-tile communication defined at compile time. We show this hardware allows us to exploit spatial and temporal locality by enabling local data transfer between cores. We evaluate the bottlenecks in this SRAM-only implementation: inter-tile bandwidth, per-tile SRAM capacity, and workload imbalance from non-uniform Gaussian density. We analyse how these constraints affect performance and render quality. This exploration raises broader questions for conventional GPUs and 3D representations, suggesting that direct inter-SM (streaming multiprocessor) communication might offer ways to reduce DRAM access in GPU kernels. We discuss these implications for the future of on-sensor and DRAM-free architectures. Project page: https://nmjfry.github.io/ipu-3dgs/
Hubert Dymarkowski, Xingjian Fu, Rappy Saha +2cs.AR cs.CV cs.DC cs.LG
Deploying Vision Transformer (ViT) models on edge platforms remains challenging due to their high computational demands and the architectural heterogeneity of modern hybrid ViT models, which incorporate both fully connected and convolutional layers. This heterogeneity leads to significant variation in tensor shapes, requiring flexible and efficient FPGA-based acceleration. In this paper, we present FlexViT, a reconfigurable FPGA accelerator for efficient ViT inference on resource-constrained edge devices. Built on the SECDA-TFLite framework, FlexViT employs a hardware-software co-design approach that maps both fully connected and convolutional layers onto a unified high-throughput INT8 GEMM engine using a runtime im2col transformation. To efficiently support diverse layer configurations, we propose a dual-mode dataflow that dynamically switches between input and weight reuse by reconfiguring the compute array at runtime. We further introduce a depth-first tiling strategy that completes accumulation in a single pass, eliminating off-chip partial-sum transfers and reducing memory bandwidth requirements. We implement FlexViT on a PYNQ-Z2 FPGA and evaluate it across a representative set of ViT models. FlexViT achieves up to 2.74x speedup on accelerator-executed layers, translating into up to 1.40x end-to-end speedup compared to CPU-only execution. The code is available at: https://github.com/gicLAB/FlexViT
Xavier Routh, Abdul Rafae Noor, Akash Kothari +4cs.PL cs.CL cs.PF
As Moore's law reaches its physical and economic limits, domain-specific approaches are increasingly employed to accelerate machine learning workloads. Hyperdimensional Computing (HDC) represents one such emerging paradigm, offering an alternative to conventional deep learning techniques. Rooted in cognitive models of computation, HDC is designed bottom-up with hardware efficiency as a first-class objective. HDC workloads map naturally to heterogeneous hardware platforms, including CPUs, GPUs, and FPGAs, as well as emerging in-memory computing technologies such as Resistive RAM (ReRAM) and Phase-Change Memory (PCM). HDC algorithms are intrinsically tolerant to noise and approximation, enabling substantial performance gains with minimal accuracy loss. In this work, we introduce ApproxHDC, a framework for automated identification and application of domain-specific approximations in HDC workloads. ApproxHDC extends the HPVM-HDC compiler infrastructure to enable retargetable compilation across diverse hardware backends, including CPUs, GPUs, and simulated ReRAM and PCM-based accelerators. The space of possible approximations is exponentially large; ApproxHDC employs efficient search and analysis to navigate it and identify high-impact configurations spanning both software and hardware levels.
Chengwei Zhou, Ovishake Sen, Xuming Chen +5eess.IV cs.CV
Deploying vision transformers (ViTs) on sensor-edge systems is limited not only by on-device compute, but also by the energy and bandwidth required to transmit high-dimensional image data from the sensor to the processor. While in-sensor and near-sensor computing reduce this cost through early feature extraction, existing methods often provide only modest compression. We observe that the frequency domain provides a naturally compact representation of visual information and can be exploited at the sensor level to reduce sensor-to-processor data movement. Building on this insight, we present FrequencyFormer, a co-designed sensor-to-processor pipeline for efficient ViT inference. FrequencyFormer includes: (1) a multi-scale DCT tokenizer that compresses a 224x224 image into compact frequency-domain tokens, achieving up to 128x reduction in off-chip data volume with modest accuracy loss; (2) a LUT-based near-sensor hardware implementation that leverages fixed DCT coefficients for multiplier-free, energy- and area-efficient tokenization; and (3) a modified MIPI-based low-power communication architecture that further reduces transfer energy. FrequencyFormer serves as a drop-in replacement for standard ViT patch embedding and remains compatible with pretrained backbones across classification, detection, and segmentation tasks. The pipeline achieves 28.8 TOPS/W, reduces communication energy by 230x, and lowers total sensor-side energy by 2.22x, demonstrating frequency-domain tokenization as a scalable foundation for in-sensor ViT deployment.
Transformer-based models achieve strong performance for jet tagging at the CERN LHC, but deploying them in low-latency, resource-constrained trigger systems is challenging. We present an initial implementation of a quantized, integer-only transformer for jet tagging on the AMD Versal AI Engine (AIE), mapping dense and multi-head attention (MHA) layers to AIE tiles. The main contribution is a reusable software framework that represents transformer layers as composable AIE building blocks and automatically generates the corresponding Vitis graph code from a high-level Python model description. This framework provides a foundation for future research and is released as open-source software at https://github.com/KastnerRG/particle_transformer_aie.
Simple linear and frequency-domain models remain surprisingly competitive in long-horizon time-series forecasting, and recent mechanistic evidence suggests that standard forecasting benchmarks may not require the dense superposed representations that make transformers powerful in other domains. This raises a substrate-level question: if the core forecasting operator is often low-complexity and approximately linear, does it need to be implemented as learned digital temporal mixing? We introduce HAMON, a passive diffractive optical forecasting core in which historical values are encoded onto an optical aperture, future positions are left dark, and cascaded trainable phase masks with free-space diffraction shape the forecast directly in the output field. At inference, prediction is performed by a single passive optical propagation pass with no trainable digital sequence-mixing layer. Across standard benchmarks, HAMON outperforms the strongest digital baselines considered on ETTm2 at all horizons and on ETTh2 at all but the longest horizon, improving MSE by up to 14\% and doing so consistently across horizons rather than at isolated points. It is competitive on Weather and trails the strongest baselines on the remaining ETT settings and on the high-channel-count Traffic and Electricity datasets. Phase encoding, intensity-compatible readout, and phase-scrambling ablations, together with a TorchOptics cross-simulator check, indicate that the forecasts arise from the data-bearing optical field rather than from a digital forecasting head. Because the passive core uses standard Fourier optics, HAMON defines a concrete target for optical hardware and for passive physical sequence mixing.
Jonas Crols, Guilherme Paim, Shirui Zhao +1cs.AR cs.LG
Bayesian Neural Networks (BNNs) offer opportunities for greatly enhancing the trustworthiness of conventional neural networks by monitoring the uncertainties in decision-making. A significant drawback for BNN inference at the extreme edge, however, is the imperative need to incorporate Gaussian Random Number Generators (GRNG) within each neuron. State-of-the-art GRNG algorithms heavily depend on multiple arithmetic operations and the use of extensive look-up tables, posing significant implementation challenges for ultra-low power hardware implementations. To overcome this, this paper presents an innovative binary tree random number generator (TreeGRNG) allowing the use of ultra-low-cost constant comparators instead of arithmetic units. We further enhance the TreeGRNG proposal with a set of hardware-aware optimizations exploiting the Gaussian properties. The optimized TreeGRNG surpasses the State-of-the-Art (SoTA) in terms of distribution accuracy while achieving a 3.7$\times$ reduction in energy per sample and boosting the throughput per unit area by 5.8$\times$. Moreover, our TreeGRNG proposal possesses a distinct advantage over the current SoTA in terms of flexibility, as it easily enables designers to adjust the shape of the sampled probability distribution, extending beyond the capabilities of traditional GRNGs, opening the horizon towards future probabilistic AI designs. The TreeGRNG design is available open-source in the link
ANEForge is a Python package that programs the Apple Neural Engine (ANE), the fixed-function neural accelerator on every recent Apple device, directly and without CoreML. In production the engine is reachable only through CoreML, which treats it as a scheduling option: no configuration requires the ANE, and a model can silently run on the CPU or GPU instead. ANEForge compiles a lazy tensor graph, built from 58 fused operators and 19 native bridge operators, into a single ANE program. The program is dispatched through the same ANE daemon and kernel-driver stack as Apple's internal framework. Beyond inference, the package reaches the engine's native fused attention, streams int8, int4, and sparse weights, keeps decoder and optimizer state resident across steps, and runs the forward pass, backward pass, and optimizer update of training on the engine. A small fused program completes a call in about 90us, near the engine's 70us per-program dispatch floor, and a pretrained ResNet-18 forward runs end-to-end in 0.33ms. ResNet-18, a sentence encoder, and a Vision Transformer run end-to-end against framework references, and a Stable Diffusion U-Net validates its forward pass. ANEForge targets Apple Silicon under macOS 14 and later. Each release is verified against a recorded macOS and ANE-compiler version.
While Ising machines serve as advanced physical solvers for the Ising model,enabling applications in combinatorial optimization and neural network training,their scalability for large-scale neural networks remains constrained by hardware connectivity limitations and suboptimal training methodologies. In this work,we leverage a Coherent Ising Machine (CIM) to train an energy-based neural network using Equilibrium Propagation, achieving performance comparable to existing software-based implementations. We further enhance the algorithm by integrating the Adam optimizer to solve for the ground state of a Hopfield energy network, significantly improving convergence speed and solution accuracy. Additionally, we demonstrate the scalability of our approach across deeper network architectures and convolutional operations. Our results highlight the potential of CIM dynamics as a scalable platform for training complex neural networks, offering a pathway toward energy-efficient implementations via analog circuits, optoelectronics, or integrated photonics. This work establishes a novel physical framework for next-generation AI hardware development.