arXiv subjects

Zhiru Zhang

Publications and source records attributed to Zhiru Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Gate the Filter, Not the Message: Node-Channel Mixtures for Pre-Propagation GNNs

Pre-propagation graph neural networks (PPGNNs) push all graph-dependent computation into a preprocessing step and train only on the resulting dense hop features, which makes them highly scalable. A puzzle in this regime is that more complex hop aggregators do not reliably outperform simpler ones: on many benchmarks, a plain MLP-based aggregator matches or beats hop-attention variants. We revisit this behavior from a graph-filter perspective. Over a precomputed diffusion basis, existing PPGNNs differ mainly in how filter coefficients are shared across nodes and feature channels, rather than simply in raw aggregator capacity. MLP-based architectures learn channel-dependent filters that are largely shared across nodes, while hop-attention-based architectures learn node-dependent mixtures that are largely shared across channels. This reveals a missing regime in standard PPGNN designs: joint node- and channel-adaptive filtering under the pre-propagation computational contract. We propose FilterMoE, a mixture-of-experts PPGNN in which a small bank of learnable Chebyshev filter experts is routed jointly over nodes and channels by a 3D gating tensor. Across eleven homophilic and heterophilic benchmarks, FilterMoE outperforms strong PPGNN baselines on nine datasets and ranks first on all three large-scale benchmarks, improving the average test score by 1.53 points. These results establish joint node-channel filter routing as a robust alternative to dataset-specific hop-aggregator selection.

cs.LG↗

arXiv2026

From Human Guidance to Autonomy: Agent Skill System for End-to-End LLM Deployment on Spatial NPUs

Spatial neural processing units (NPUs) provide an energy-efficient platform for edge LLM inference, but efficiently deploying an LLM end-to-end on such hardware remains labor-intensive. Although AI coding agents have begun to lower this cost, existing studies have largely focused on single-kernel optimization rather than end-to-end LLM deployment on resource-constrained spatial NPUs. We present a two-stage methodology, instantiated on the AMD XDNA 2 NPU, that progresses from human-guided development to agent autonomy. In the first stage, we develop a reference deployment of Llama-3.2-1B through human-guided agent assistance. The resulting implementation achieves a speedup of 2.2x on prefill and 4.0x on decode over the hand-optimized baseline, with the optimization trajectory and its lessons recorded as structured documentation throughout. In the second stage, we distill the documentation into an agent skill system consisting of eight phases, orchestrating the optimization and debugging skill sets, with numerical correctness strictly enforced at each phase. Using our agent skill system, we autonomously deploy eight additional decoder-only LLMs (Llama-3.2-3B, SmolLM2-1.7B, Qwen2.5-{0.5B, 1.5B, 3B}, Qwen3-{0.6B, 1.7B, 4B}) end-to-end on the AMD XDNA 2 NPU using the open-source compiler stack. To our knowledge, these models have not previously been deployed on AMD NPUs via any open-source software stack. Each deployment completes in 0.5-4 hours of agent wall time with almost no human guidance, and passes the numerical-correctness gates, demonstrating functional generalization to previously unencountered LLMs. Three of the eight match or exceed the sustained performance of our Llama-3.2-1B reference deployment, suggesting that the resulting implementations can be competitive without additional model-specific human engineering.

cs.LG↗

arXiv2026

Revisiting Pre-Propagation GNNs: Robust Diffusion Operators and Hidden-State Re-Propagation

Pre-propagation graph neural networks (PPGNNs) decouple node feature propagation from transformation: graph diffusion is performed once as preprocessing, and training reduces to dense per-node transformations. This design enables mini-batch training without inter-node dependencies, avoids repeated sparse matrix--matrix multiplications, and better matches modern accelerators optimized for dense compute. However, their expressivity remains unclear, and empirical results show a gap between PPGNNs and their message-passing counterparts on commonly used graph benchmarks, especially heterophilic ones. In this paper, we propose a suite of robust graph diffusion operators for preprocessing and a few-shot hidden-state re-propagation scheme during training. Our methods improve the validation and test accuracy of PPGNNs, enabling them to match the accuracy of message-passing GNNs while maintaining training efficiency.

cs.LG↗

arXiv2026

AsyncSparse: Accelerating Sparse Matrix-Matrix Multiplication on Asynchronous GPU Architectures

Sparse Matrix-Matrix Multiplication (SpMM) is a fundamental kernel across scientific computing and machine learning. While prior work accelerates SpMM using Tensor Cores, no existing sparse kernel exploits the asynchronous features of modern GPU architectures, such as NVIDIA's Tensor Memory Accelerator (TMA) and warp specialization. This work systematically studies how these features impact SpMM performance and introduces two co-designed kernels. For structured sparsity, we optimize a warp-specialized producer-consumer pipeline overlapping TMA data transfer with WGMMA computation using Block Compressed Sparse Row (BCSR) format. For irregular sparsity, we design a Window Compressed Sparse Row (WCSR) kernel that loads the sparse operand via TMA and splits large row-windows across thread blocks for load balancing. Our WCSR kernel outperforms all prior SpMM kernels on SuiteSparse matrices (1.47x over AccSpMM, 6.24x over cuSPARSE). Our BCSR kernel achieves a combined 2.66x end-to-end speedup on Qwen2.5-7B prefill at 90% block sparsity with 64K tokens over cuDNN/cuBLAS.

cs.DC↗

arXiv2026

GauS: Differentiable Scheduling Optimization via Gaussian Reparameterization

Efficient operator scheduling is a fundamental challenge in software compilation and hardware synthesis. While recent differentiable approaches have sought to replace traditional ones like exact solvers or heuristics with gradient-based search, they typically rely on categorical distributions that fail to capture the ordinal nature of time and suffer from a parameter space that scales poorly. In this paper, we propose a novel differentiable framework, GauS, that models operator scheduling as a stochastic relaxation using Gaussian distributions, which fully utilize modern parallel computing devices like GPUs. By representing schedules as continuous Gaussian variables, we successfully capture the ordinal nature of time and reduce the optimization space by orders of magnitude. Our method is highly flexible to represent various objectives and constraints, which provides the first differentiable formulation for the complex pipelined scheduling problem. We evaluate our method on a range of benchmarks, demonstrating that Gaus achieves Pareto-optimal results.

cs.LG↗

arXiv2026

From Pragmas to Partners: A Symbiotic Evolution of Agentic High-Level Synthesis

The rise of large language models has sparked interest in AI-driven hardware design, raising the question: does high-level synthesis (HLS) still matter in the agentic era? We argue that HLS remains essential. While we expect mature agentic hardware systems to leverage both HLS and RTL, this paper focuses on HLS and its role in enabling agentic optimization. HLS offers faster iteration cycles, portability, and design permutability that make it a natural layer for agentic optimization. This position paper makes three contributions. First, we explain why HLS serves as a practical abstraction layer and a golden reference for agentic hardware design. Second, we identify key limitations of current HLS tools, namely inadequate performance feedback, rigid interfaces, and limited debuggability that agents are uniquely positioned to address. Third, we propose a taxonomy for the symbiotic evolution of agentic HLS, clarifying how responsibility shifts from human designers to AI agents as systems advance from copilots to autonomous design partners.

cs.CL↗

arXiv2026

Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve

Modern compilers rely on hand-crafted heuristics to guide optimization passes. These human-designed rules often struggle to adapt to the complexity of modern software and hardware and lead to high maintenance burden. To address this challenge, we present Magellan, an agentic framework that evolves the compiler pass itself by synthesizing executable C++ decision logic. Magellan couples an LLM coding agent with evolutionary search and autotuning in a closed loop of generation, evaluation on user-provided macro-benchmarks, and refinement, producing compact heuristics that integrate directly into existing compilers. Across several production optimization tasks, Magellan discovers policies that match or surpass expert baselines. In LLVM function inlining, Magellan synthesizes new heuristics that outperform decades of manual engineering for both binary-size reduction and end-to-end performance. In register allocation, it learns a concise priority rule for live-range processing that matches intricate human-designed policies on a large-scale workload. We also report preliminary results on XLA problems, demonstrating portability beyond LLVM with reduced engineering effort.

cs.AI↗

arXiv2025

Tawa: Automatic Warp Specialization for Modern GPUs with Asynchronous References

Modern GPUs feature specialized hardware units that enable high-performance, asynchronous dataflow execution. However, the conventional SIMT programming model is fundamentally misaligned with this task-parallel hardware, creating a significant programmability gap. While hardware-level warp specialization is the key to unlocking peak performance, it forces developers to manually orchestrate complex, low-level communication and software pipelines--a process that is labor-intensive, error-prone, and unsustainable. To address this challenge, we present Tawa, an automated compiler that systematically generates high-performance, warp-specialized code from a high-level, tile-based program. Central to our approach is a novel IR abstraction, asynchronous references (aref), which expresses warp-level communication without exposing low-level hardware details. Using this abstraction, Tawa automatically partitions programs into producer-consumer roles and manages the intricate dataflow pipeline, relieving developers of invasive kernel rewriting. Evaluation on NVIDIA H100 GPUs across representative LLM kernels shows that Tawa delivers high hardware utilization, achieving up to 1.1$\times$ speedup over highly optimized cuBLAS GEMM kernels. For attention workloads, Tawa attains 1.2$\times$ speedup over Triton and matches the performance of the hand-optimized CUTLASS C++ FlashAttention-3 kernel with far less programming effort.

cs.LG↗

arXiv2025

Dato: A Task-Based Programming Model for Dataflow Accelerators

Recent deep learning workloads increasingly push computational demand beyond what current memory systems can sustain, with many kernels stalling on data movement rather than computation. While modern dataflow accelerators incorporate on-chip streaming to mitigate off-chip bandwidth limitations, existing programming models struggle to harness these capabilities effectively. Low-level interfaces provide fine-grained control but impose significant development overhead, whereas high-level tile-based languages abstract away communication details, restricting optimization and forcing compilers to reconstruct the intended dataflow. We present Dato, a Python-embedded, task-based programming model for dataflow accelerators that elevates data communication and sharding to first-class type constructs. Developers write programs as a graph of tasks connected via explicit stream types, with sharded inputs specified using layout types. These tasks are first mapped virtually onto the accelerator's spatial fabric, and the compiler then generates a physical mapping that respects hardware constraints. Experimental results on both AMD Ryzen AI NPU and Alveo FPGA devices demonstrate that Dato achieves high performance while significantly reducing the burden of writing optimized code. On the NPU, Dato attains up to 84% hardware utilization for GEMM and delivers a 2.81x speedup on attention kernels compared to a state-of-the-art commercial framework. On the FPGA, Dato surpasses leading frameworks in performance when generating custom systolic arrays, achieving 98% of the theoretical peak performance.

cs.PL↗

arXiv2025

Characterizing and Optimizing Realistic Workloads on a Commercial Compute-in-SRAM Device

Compute-in-SRAM architectures offer a promising approach to achieving higher performance and energy efficiency across a range of data-intensive applications. However, prior evaluations have largely relied on simulators or small prototypes, limiting the understanding of their real-world potential. In this work, we present a comprehensive performance and energy characterization of a commercial compute-in-SRAM device, the GSI APU, under realistic workloads. We compare the GSI APU against established architectures, including CPUs and GPUs, to quantify its energy efficiency and performance potential. We introduce an analytical framework for general-purpose compute-in-SRAM devices that reveals fundamental optimization principles by modeling performance trade-offs, thereby guiding program optimizations. Exploiting the fine-grained parallelism of tightly integrated memory-compute architectures requires careful data management. We address this by proposing three optimizations: communication-aware reduction mapping, coalesced DMA, and broadcast-friendly data layouts. When applied to retrieval-augmented generation (RAG) over large corpora (10GB--200GB), these optimizations enable our compute-in-SRAM system to accelerate retrieval by 4.8$\times$--6.6$\times$ over an optimized CPU baseline, improving end-to-end RAG latency by 1.1$\times$--1.8$\times$. The shared off-chip memory bandwidth is modeled using a simulated HBM, while all other components are measured on the real compute-in-SRAM device. Critically, this system matches the performance of an NVIDIA A6000 GPU for RAG while being significantly more energy-efficient (54.4$\times$-117.9$\times$ reduction). These findings validate the viability of compute-in-SRAM for complex, real-world applications and provide guidance for advancing the technology.

cs.AR↗

arXiv2025

e-boost: Boosted E-Graph Extraction with Adaptive Heuristics and Exact Solving

E-graphs have attracted growing interest in many fields, particularly in logic synthesis and formal verification. E-graph extraction is a challenging NP-hard combinatorial optimization problem. It requires identifying optimal terms from exponentially many equivalent expressions, serving as the primary performance bottleneck in e-graph based optimization tasks. However, traditional extraction methods face a critical trade-off: heuristic approaches offer speed but sacrifice optimality, while exact methods provide optimal solutions but face prohibitive computational costs on practical problems. We present e-boost, a novel framework that bridges this gap through three key innovations: (1) parallelized heuristic extraction that leverages weak data dependence to compute DAG costs concurrently, enabling efficient multi-threaded performance without sacrificing extraction quality; (2) adaptive search space pruning that employs a parameterized threshold mechanism to retain only promising candidates, dramatically reducing the solution space while preserving near-optimal solutions; and (3) initialized exact solving that formulates the reduced problem as an Integer Linear Program with warm-start capabilities, guiding solvers toward high-quality solutions faster. Across the diverse benchmarks in formal verification and logic synthesis fields, e-boost demonstrates 558x runtime speedup over traditional exact approaches (ILP) and 19.04% performance improvement over the state-of-the-art extraction framework (SmoothE). In realistic logic synthesis tasks, e-boost produces 7.6% and 8.1% area improvements compared to conventional synthesis tools with two different technology mapping libraries. e-boost is available at https://github.com/Yu-Maryland/e-boost.

cs.AI↗

arXiv2025

HeuriGym: An Agentic Benchmark for LLM-Crafted Heuristics in Combinatorial Optimization

While Large Language Models (LLMs) have demonstrated significant advancements in reasoning and agent-based problem-solving, current evaluation methodologies fail to adequately assess their capabilities: existing benchmarks either rely on closed-ended questions prone to saturation and memorization, or subjective comparisons that lack consistency and rigor. In this work, we introduce HeuriGym, an agentic framework designed for evaluating heuristic algorithms generated by LLMs for combinatorial optimization problems, characterized by clearly defined objectives and expansive solution spaces. HeuriGym empowers LLMs to propose heuristics, receive evaluative feedback via code execution, and iteratively refine their solutions. We evaluate nine state-of-the-art models on nine problems across domains such as computer systems, logistics, and biology, exposing persistent limitations in tool use, planning, and adaptive reasoning. To quantify performance, we propose the Quality-Yield Index (QYI), a metric that captures both solution pass rate and quality. Even top models like GPT-o4-mini-high and Gemini-2.5-Pro attain QYI scores of only 0.6, well below the expert baseline of 1. Our open-source benchmark aims to guide the development of LLMs toward more effective and realistic problem-solving in scientific and engineering domains.

cs.LG↗

arXiv2025

FlashDLM: Accelerating Diffusion Language Model Inference via Efficient KV Caching and Guided Diffusion

Diffusion language models offer parallel token generation and inherent bidirectionality, promising more efficient and powerful sequence modeling compared to autoregressive approaches. However, state-of-the-art diffusion models (e.g., Dream 7B, LLaDA 8B) suffer from slow inference. While they match the quality of similarly sized autoregressive (AR) models (e.g., Qwen2.5 7B, Llama3 8B), their iterative denoising requires multiple full-sequence forward passes, resulting in high computational costs and latency, particularly for long input prompts and long-context scenarios. Furthermore, parallel token generation introduces token incoherence problems, and current sampling heuristics suffer from significant quality drops with decreasing denoising steps. We address these limitations with two training-free techniques. First, we propose FreeCache, a Key-Value (KV) approximation caching technique that reuses stable KV projections across denoising steps, effectively reducing the computational cost of DLM inference. Second, we introduce Guided Diffusion, a training-free method that uses a lightweight pretrained autoregressive model to supervise token unmasking, dramatically reducing the total number of denoising iterations without sacrificing quality. We conduct extensive evaluations on open-source reasoning benchmarks, and our combined methods deliver an average of 12.14x end-to-end speedup across various tasks with negligible accuracy degradation. For the first time, diffusion language models achieve a comparable and even faster latency as the widely adopted autoregressive models. Our work successfully paved the way for scaling up the diffusion language model to a broader scope of applications across different domains.

cs.CL↗

arXiv2025

Semantic Compression of 3D Objects for Open and Collaborative Virtual Worlds

Traditional methods for 3D object compression operate only on structural information within the object vertices, polygons, and textures. These methods are effective at compression rates up to 10x for standard object sizes but quickly deteriorate at higher compression rates with texture artifacts, low-polygon counts, and mesh gaps. In contrast, semantic compression ignores structural information and operates directly on the core concepts to push to extreme levels of compression. In addition, it uses natural language as its storage format, which makes it natively human-readable and a natural fit for emerging applications built around large-scale, collaborative projects within augmented and virtual reality. It deprioritizes structural information like location, size, and orientation and predicts the missing information with state-of-the-art deep generative models. In this work, we construct a pipeline for 3D semantic compression from public generative models and explore the quality-compression frontier for 3D object compression. We apply this pipeline to achieve rates as high as 105x for 3D objects taken from the Objaverse dataset and show that semantic compression can outperform traditional methods in the important quality-preserving region around 100x compression.

cs.CV↗

arXiv2025

Graph Learning at Scale: Characterizing and Optimizing Pre-Propagation GNNs

Graph neural networks (GNNs) are widely used for learning node embeddings in graphs, typically adopting a message-passing scheme. This approach, however, leads to the neighbor explosion problem, with exponentially growing computational and memory demands as layers increase. Graph sampling has become the predominant method for scaling GNNs to large graphs, mitigating but not fully solving the issue. Pre-propagation GNNs (PP-GNNs) represent a new class of models that decouple feature propagation from training through pre-processing, addressing neighbor explosion in theory. Yet, their practical advantages and system-level optimizations remain underexplored. This paper provides a comprehensive characterization of PP-GNNs, comparing them with graph-sampling-based methods in training efficiency, scalability, and accuracy. While PP-GNNs achieve comparable accuracy, we identify data loading as the key bottleneck for training efficiency and input expansion as a major scalability challenge. To address these issues, we propose optimized data loading schemes and tailored training methods that improve PP-GNN training throughput by an average of 15$\times$ over the PP-GNN baselines, with speedup of up to 2 orders of magnitude compared to sampling-based GNNs on large graph benchmarks. Our implementation is publicly available at https://github.com/cornell-zhang/preprop-gnn.

cs.LG↗

arXiv2024

RapidStream IR: Infrastructure for FPGA High-Level Physical Synthesis

The increasing complexity of large-scale FPGA accelerators poses significant challenges in achieving high performance while maintaining design productivity. High-level synthesis (HLS) has been adopted as a solution, but the mismatch between the high-level description and the physical layout often leads to suboptimal operating frequency. Although existing proposals for high-level physical synthesis, which use coarse-grained design partitioning, floorplanning, and pipelining to improve frequency, have gained traction, they lack a framework enabling (1) pipelining of real-world designs at arbitrary hierarchical levels, (2) integration of HLS blocks, vendor IPs, and handcrafted RTL designs, (3) portability to emerging new target FPGA devices, and (4) extensibility for the easy implementation of new design optimization tools. We present RapidStream IR, a practical high-level physical synthesis (HLPS) infrastructure for representing the composition of complex FPGA designs and exploring physical optimizations. Our approach introduces a flexible intermediate representation (IR) that captures interconnection protocols at arbitrary hierarchical levels, coarse-grained pipelining, and spatial information, enabling the creation of reusable passes for design frequency optimizations. RapidStream IR improves the frequency of a broad set of mixed-source designs by 7% to 62%, including large language models and genomics accelerators, and is portable to user-customizable new FPGA platforms. We further demonstrate its extensibility through case studies, showcasing the ability to facilitate future research.

cs.AR↗

arXiv2024

Rapid GPU-Based Pangenome Graph Layout

Computational Pangenomics is an emerging field that studies genetic variation using a graph structure encompassing multiple genomes. Visualizing pangenome graphs is vital for understanding genome diversity. Yet, handling large graphs can be challenging due to the high computational demands of the graph layout process. In this work, we conduct a thorough performance characterization of a state-of-the-art pangenome graph layout algorithm, revealing significant data-level parallelism, which makes GPUs a promising option for compute acceleration. However, irregular data access and the algorithm's memory-bound nature present significant hurdles. To overcome these challenges, we develop a solution implementing three key optimizations: a cache-friendly data layout, coalesced random states, and warp merging. Additionally, we propose a quantitative metric for scalable evaluation of pangenome layout quality. Evaluated on 24 human whole-chromosome pangenomes, our GPU-based solution achieves a 57.3x speedup over the state-of-the-art multithreaded CPU baseline without layout quality loss, reducing execution time from hours to minutes.

cs.DC↗

arXiv2024

ShadowLLM: Predictor-based Contextual Sparsity for Large Language Models

The high power consumption and latency-sensitive deployments of large language models (LLMs) have motivated efficiency techniques like quantization and sparsity. Contextual sparsity, where the sparsity pattern is input-dependent, is crucial in LLMs because the permanent removal of attention heads or neurons from LLMs can significantly degrade accuracy. Prior work has attempted to model contextual sparsity using neural networks trained to predict activation magnitudes, which can be used to dynamically prune structures with low predicted activation magnitude. In this paper, we look beyond magnitude-based pruning criteria to assess attention head and neuron importance in LLMs. We develop a novel predictor called ShadowLLM, which can shadow the LLM behavior and enforce better sparsity patterns, resulting in over 15% improvement in end-to-end accuracy compared to prior methods. In addition, ShadowLLM achieves up to a 20% speed-up over the state-of-the-art DejaVu framework. These enhancements are validated on Llama-2 and OPT models with up to 30 billion parameters. Our code is available at \href{https://github.com/abdelfattah-lab/shadow_llm/}{ShadowLLM}.

cs.LG↗