Search: “hardware”

arXiv:2605.19269v1 Announce Type: new Abstract: Transformer training systems are built around dense linear algebra, yet a nontrivial fraction of end-to-end time is spent on surrounding memory-bound op

arXiv:2603.09473v2 Announce Type: replace Abstract: Equipping robotic systems with the capacity to generate extit{ex novo} hardware during operation extends control of physical adaptability. Unlike mo

arXiv:2605.18889v1 Announce Type: cross Abstract: Modern machine learning forces practitioners to choose between powerful but expensive deep networks and fast but limited classical algorithms. Here we

arXiv:2605.17164v1 Announce Type: cross Abstract: Deploying large-scale LLM training and inference with optimal performance is exceptionally challenging due to a complex design space of parallelism st

arXiv:2602.06694v2 Announce Type: replace Abstract: Weight-only quantization has become a standard approach for efficiently serving large language models (LLMs). However, existing methods fail to effi

arXiv:2605.17991v1 Announce Type: cross Abstract: Stable Audio 3 is a family of fast latent diffusion models (small, medium, large) for variable-length audio generation and editing. Since our models c

arXiv:2605.15237v1 Announce Type: cross Abstract: Accelerating applications through the design of hardware accelerators can significantly enhance system performance and energy efficiency. Despite adva

arXiv:2605.14624v1 Announce Type: cross Abstract: A common critique of neural combinatorial-optimization solvers is that they are less energy-efficient than CPU metaheuristics, given the operational e

arXiv:2605.14665v1 Announce Type: new Abstract: Legal reasoning is not semantic similarity search. A court judgment encodes constrained symbolic reasoning: precedent propagation, procedural state tran

arXiv:2506.04499v2 Announce Type: replace Abstract: Existing LiDAR 3D object detection methods predominantely rely on sparse convolutions and/or transformers, which can be challenging to run on resour

arXiv:2605.13925v1 Announce Type: new Abstract: Robotic dexterous hands are central to contact-rich manipulation, with rapid progress driven by advances in hardware, sensing, control, simulation, and

arXiv:2605.14844v1 Announce Type: cross Abstract: We introduce XFP, a dynamic weight quantizer for LLM inference that inverts the conventional workflow: the operator specifies reconstruction quality f

arXiv:2509.19929v4 Announce Type: replace-cross Abstract: Uncertainty Quantification (UQ) is paramount for inference in engineering. A common inference task is to recover full-field information of phy

arXiv:2605.11333v1 Announce Type: cross Abstract: The fast pace of artificial intelligence~(AI) innovation demands an agile methodology for observation, reproduction and optimization of distributed ma

arXiv:2605.10457v1 Announce Type: cross Abstract: Real-time Light Detection And Ranging (LiDAR) simulation must find, per emitted ray, the closest intersecting triangle even in dynamic scenes containi

arXiv:2605.10886v1 Announce Type: cross Abstract: Recent GPU generations deliver significantly higher FLOPs using lower-precision arithmetic, such as FP8. While successfully applied to large language

arXiv:2605.08913v1 Announce Type: cross Abstract: Autoregressive inference is typically assumed to scale predictably with decoding length, and key-value (KV) caching is widely regarded as a universall

arXiv:2603.29002v2 Announce Type: replace-cross Abstract: Modern large language models (LLMs) increasingly depends on efficient long-context processing and generation mechanisms, including sparse atte

arXiv:2603.02883v3 Announce Type: replace Abstract: Diffusion Transformers (DiTs) achieve state-of-the-art video generation quality, but their substantial memory and computational footprints hinder ed

arXiv:2605.06763v1 Announce Type: new Abstract: Sparse attention improves LLM inference efficiency by selecting a subset of key-value entries, but at the cost of potential accuracy degradation. In par

arXiv:2509.05276v4 Announce Type: replace-cross Abstract: Mainstream Transformer-based large language models face major efficiency bottlenecks: training computation scales quadratically with sequence

arXiv:2605.06927v1 Announce Type: cross Abstract: Object detection on heterogeneous edge devices must satisfy strict energy, latency, and memory constraints while still providing reliable perception f

arXiv:2605.04357v1 Announce Type: cross Abstract: The usage of large language models (LLMs) has grown increasingly fragmented, with no single model dominating. Meanwhile, cloud providers offer a wide

arXiv:2605.05049v1 Announce Type: cross Abstract: Frontier models increasingly adopt Mixture-of-Experts (MoE) architectures to achieve large-model performance at reduced cost. However, training MoE mo

arXiv:2605.01352v1 Announce Type: cross Abstract: GPU-based simulation environments for embodied AI interleave physics simulation (CUDA) and photorealistic rendering (Vulkan) on a single device. We ob

arXiv:2603.18066v2 Announce Type: replace-cross Abstract: Backpropagation has enabled modern deep learning but is difficult to realize as an online, fully distributed hardware learning system due to g

arXiv:2602.07264v2 Announce Type: replace Abstract: Unmanned aerial vehicles are rapidly transforming multiple applications, from agricultural and infrastructure monitoring to logistics and defense. I

arXiv:2602.02958v4 Announce Type: replace Abstract: Despite rapid progress in autoregressive video diffusion, an emerging system algorithm bottleneck limits both deployability and generation capabilit

arXiv:2605.00174v1 Announce Type: cross Abstract: Video and image streaming on edge devices requires low latency. To address this, Neural Networks (NNs) are widely used, and prior work mainly focuses

arXiv:2604.27085v1 Announce Type: cross Abstract: Fine-tuning Large Language Models (LLMs) on consumer-grade GPUs is highly cost-effective, yet constrained by limited GPU memory and slow PCIe intercon

arXiv:2604.26968v1 Announce Type: cross Abstract: Key-value (KV) cache memory management is the primary bottleneck limiting throughput and cost-efficiency in large-scale GPU inference serving. Current

arXiv:2604.25326v2 Announce Type: replace-cross Abstract: Speculative decoding enhances the inference efficiency of large language models (LLMs) by generating drafts using a small draft language model

arXiv:2604.26103v1 Announce Type: cross Abstract: All current LLM serving systems place the GPU at the center, from production-level attention-FFN disaggregation to NVIDIA's Rubin GPU-LPU heterogeneou

arXiv:2603.15954v2 Announce Type: replace Abstract: Real-time AI experiences call for on-device large language models (OD-LLMs) optimized for efficient deployment on resource-constrained hardware. The

arXiv:2604.24169v1 Announce Type: new Abstract: 3D point cloud perception remains tightly coupled to custom CUDA operators for spatial operations, limiting portability and efficiency on non-NVIDIA, AM

arXiv:2505.02922v3 Announce Type: replace Abstract: Recent large language models (LLMs) are rapidly extending their context windows, yet inference throughput lags due to increasing GPU memory and band

arXiv:2604.21743v1 Announce Type: new Abstract: Image enhancement models for mobile devices often struggle to balance high output quality with the fast processing speeds required by mobile hardware. W

arXiv:2604.16395v2 Announce Type: replace-cross Abstract: Context retrieval systems for LLM inference face a critical challenge: high retrieval latency creates a fundamental tension between waiting fo

arXiv:2412.03594v3 Announce Type: replace-cross Abstract: Large language models (LLMs) increasingly play an important role in a wide range of information processing and management tasks in industry. M

arXiv:2604.20819v1 Announce Type: new Abstract: The scalability of long-context large language models is fundamentally limited by the quadratic memory cost of exact self-attention, which often leads t

arXiv:2604.18616v1 Announce Type: cross Abstract: LLM-based coding agents can generate functionally correct GPU kernels, yet their performance remains far below hand-optimized libraries on critical co

arXiv:2412.00069v3 Announce Type: replace-cross Abstract: Mixture-of-Experts (MoE) has garnered significant attention for its ability to scale up neural networks while utilizing the same or even fewer

arXiv:2604.17808v1 Announce Type: cross Abstract: Zero-knowledge proof (ZKP) provers remain costly because multi-scalar multiplication (MSM) and number-theoretic transforms (NTTs) dominate runtime as

arXiv:2602.11092v2 Announce Type: replace Abstract: Identifying where quantum models may offer practical benefits in near term quantum machine learning (QML) requires moving beyond isolated algorithmi

arXiv:2510.04008v5 Announce Type: replace Abstract: Softmax Attention has a quadratic time complexity in sequence length, which becomes prohibitive to run at long contexts, even with highly optimized

arXiv:2604.16715v1 Announce Type: cross Abstract: Graph foundation models have demonstrated remarkable adaptability across diverse downstream tasks through large-scale pretraining on graphs. However,