Arghadip Das

Hoseok Kim, Arghadip Das, Soumendu Ghosh et al.

Modern edge System-on-Chips (SoCs) integrate heterogeneous processing units (PUs) such as CPUs, GPUs, and NPUs, yet current inference stacks map entire models to a single PU, leaving significant performance and energy efficiency on the table. This is exacerbated by emerging architectures such as state-space models (SSMs), Kolmogorov-Arnold networks (KANs), and multi-stage vision-language-action (VLA) pipelines, whose diverse operator characteristics are not uniformly suited to any single PU. We present BIDENT, a unified operator-level orchestration framework for heterogeneous edge inference that maps individual operators to the most suitable PU based on profiled execution characteristics. BIDENT formulates operator-to-PU assignment as a shortest-path problem over a weighted execution graph, enabling efficient and optimal scheduling under the cost model for both latency- and energy-minimization objectives. Unlike prior work relying on model-specific heuristics or coarse-grained partitioning, BIDENT is model-agnostic and jointly supports sequential execution, intra-model parallelism across independent operators, and multi-model concurrent scheduling in a single formulation. We implement BIDENT on an Intel Core Ultra SoC and evaluate it across 10 model families spanning CNNs, Transformers, SSMs, KANs, spiking networks, and multi-stage pipelines. BIDENT achieves up to 1.60x speedup via intra-model parallelism and a 3.42x geometric mean speedup across 190 multi-model combinations by utilizing otherwise idle compute. Sequential heterogeneous mapping yields more modest gains (up to 1.58x, 1.09x geometric mean), while energy-aware scheduling reduces energy consumption by 48.2% on average in concurrent settings. These results show that operator-level orchestration, not model-level mapping, is the key abstraction for fully exploiting heterogeneity in next-generation edge AI.

Arghadip Das, Hoseok Kim, Soomin Lee et al.

AI model architectures are diversifying rapidly. Although dense matrix multiplication underlies today's CNNs and transformers, emerging architectures (state-space models, long convolutions via the fast Fourier transform (FFT), Kolmogorov-Arnold networks, and spiking networks) are not multiply-accumulate (MAC) dominated; they spend much of their computation on vector and non-MAC primitives that homogeneous, MAC-centric neural processing units (NPUs) serve poorly. This has motivated heterogeneous NPUs (HPUs) built from non-identical tiles. Prior heterogeneous designs vary only one or two coarse knobs (typically MAC precision or array size) and are evaluated on narrow workloads; no existing framework supports fine-grained HPU design, where tiles differ across many architectural dimensions at once. We present MOSAIC, an analytical simulator and design-space-exploration (DSE) framework for HPU microarchitecture design. MOSAIC searches the joint space of tile-level heterogeneity: beyond array size and precision, it varies tile-type composition (large Big, small Little, and non-MAC Special-Function tiles), dataflow, sparsity mode, MAC engine type, and special-function units for non-MAC operators (FFT, spiking-integrate, polynomial). Unlike prior simulators that model a single homogeneous tile type, MOSAIC models non-MAC tiles with their own energy, area, and timing models and maps operators across a mix of tiles with a heterogeneity-aware compiler. A multi-seed pipeline pairing a stratified sweep with genetic-algorithm refinement returns Pareto-optimal designs, with cost models calibrated to a 7 nm node and cross-validated against NVIDIA's Deep Learning Accelerator (NVDLA). Across a 20-workload suite, the best general-purpose HPU found by MOSAIC (~200 mm^2 Big+Little+Special-Function) achieves +46.91% mean iso-area energy savings over the best iso-area homogeneous baseline.

Arghadip Das, Arnab Raha, Shamik Kundu et al.

State-Space Models (SSMs) have emerged as efficient alternatives to transformers for sequential data tasks, offering linear or near-linear scalability with sequence length, making them ideal for long-sequence applications in NLP, vision, and edge AI, including real-time transcription, translation, and contextual search. These applications require lightweight, high-performance models for deployment on resource-constrained devices like laptops and PCs. Designing specialized accelerators for every emerging neural network is costly and impractical; instead, optimizing models for existing NPUs in AI PCs provides a scalable solution. To this end, we propose XAMBA, the first framework to enable and optimize SSMs on commercial off-the-shelf (COTS) state-of-the-art (SOTA) NPUs. XAMBA follows a three-step methodology: (1) enabling SSMs on NPUs, (2) optimizing performance to meet KPI requirements, and (3) trading accuracy for additional performance gains. After enabling SSMs on NPUs, XAMBA mitigates key bottlenecks using CumBA and ReduBA, replacing sequential CumSum and ReduceSum operations with matrix-based computations, significantly improving execution speed and memory efficiency. Additionally, ActiBA enhances performance by approximating expensive activation functions (e.g., Swish, Softplus) using piecewise linear mappings, reducing latency with minimal accuracy loss. Evaluations on an Intel Core Ultra Series 2 AI PC show that XAMBA achieves up to 4.8X speed-up over the baseline. Our implementation is available at https://github.com/arghadippurdue/XAMBA.

Arghadip Das, Shamik Kundu, Arnab Raha et al.

Graph Neural Networks (GNNs) are vital for learning from graph-structured data, enabling applications in network analysis, recommendation systems, and speech analytics. Deploying them on edge devices like client PCs and laptops enhances real-time processing, privacy, and cloud independence. GNNs aid Retrieval-Augmented Generation (RAG) for Large Language Models (LLMs) and enable event-based vision tasks. However, irregular memory access, sparsity, and dynamic structures cause high latency and energy overhead on resource-constrained devices. While modern edge processors integrate CPUs, GPUs, and NPUs, NPUs designed for data-parallel tasks struggle with irregular GNN computations. We introduce GraNNite, the first hardware-aware framework optimizing GNN execution on commercial-off-the-shelf (COTS) SOTA DNN accelerators via a structured three-step methodology: (1) enabling NPU execution, (2) optimizing performance, and (3) trading accuracy for efficiency gains. Step 1 employs GraphSplit for workload distribution and StaGr for static aggregation, while GrAd and NodePad handle dynamic graphs. Step 2 boosts performance using EffOp for control-heavy tasks and GraSp for sparsity exploitation. Graph Convolution optimizations PreG, SymG, and CacheG reduce redundancy and memory transfers. Step 3 balances quality versus efficiency, where QuantGr applies INT8 quantization, and GrAx1, GrAx2, and GrAx3 accelerate attention, broadcast-add, and SAGE-max aggregation. On Intel Core Ultra AI PCs, GraNNite achieves 2.6X to 7.6X speedups over default NPU mappings and up to 8.6X energy gains over CPUs and GPUs, delivering 10.8X and 6.7X higher performance than CPUs and GPUs, respectively, across GNN models.