arXiv subjects

Mengming Li

Publications and source records attributed to Mengming Li.

16 recordsLinked to original sources

arXiv2026

Memory Compression for High-Fanout Agent Sandboxes

High-fanout agent workloads create a growing memory bottleneck because a single task may spawn many concurrent sandbox sessions. Yet these sandboxes are far from independent: they originate from a shared template and execute related trajectories, exposing substantial template-relative and cross-sandbox memory redundancy. Conventional memory compression is poorly matched to this setting in three fundamental dimensions: how to compress, because they fail to exploit similarity across non-identical sandbox pages; what to compress, because they control page-fault overhead through conservative page selection; and when to compress, because compression is either triggered by memory pressure or performed without awareness of agent execution phases. We present AgentZip, the first memory compression system designed specifically for AI-agent sandboxes. AgentZip introduces compression mechanisms that exploit both the template-relative and cross-sandbox redundancy. It broadens the compression scope to any page with a profitable representation and shifts overhead control from compression-time page selection to restore-time prefetching. It further aligns expensive compression with LLM waiting periods to avoid interfering with foreground tool execution. Across LLM training and inference workloads, AgentZip reduces sandbox-owned memory by up to 8.7x, compared with 2.1x for the Linux configuration. Restore prefetching and agent-execution-aware scheduling reduce the slowdown of aggressive compression from as high as 3.1x to 1.40x while retaining nearly all of its memory-saving benefit.

cs.AI↗

arXiv2026

G-Power: Architecture-level GPU Power Modeling with Aggregated Knowledge Foundations from Known GPUs

Graphics Processing Units (GPUs) have been serving as critical computation resources for large-scale parallel computations. With increasing chip complexity, power efficiency has become an important design objective for modern GPUs. GPU power optimization relies on fast power evaluation, requiring architecture-level GPU power model. However, because of the time-consuming power label collection, only simple microbenchmarks are adopted for training. The limitation of microbenchmarks as training data incurs low accuracy for existing architecture-level GPU power models like AccelWattch. To address the limitation of microbenchmarks as training data, we propose G-Power, an architecture-level GPU power modeling framework that utilizes additional known GPU chips to provide additional knowledge. G-Power utilizes the aggregated knowledge foundation from additional known GPU chips and then performs fine-tuning on our target GPU. To provide foundations with additional known GPU chips and capture the similarity to utilize these foundations for fine-tuning, G-Power adopts a three-phase algorithm consisting of 1) pre-training with additional known chips, 2) attention-inspired aggregation, and 3) fine-tuning on our target GPU. We evaluate G-Power on four modern NVIDIA GPUs, demonstrating high accuracy. G-Power can achieve a low MAPE of 14% and a high correlation coefficient R of 0.88 on average, which are 22% lower MAPE and 0.36 higher R than AccelWattch.

cs.AR↗

arXiv2026

LEAP: A Self-Supervised Per-Cycle Toggle Propagation Model Supports Fast, Transferable, and Early Analysis of Layout Power

Accurate power analysis is critical in VLSI design, as it directly impacts power optimization strategies. However, traditional approaches are often hindered by the substantial runtime required for per-cycle toggle propagation in the netlist, which propagates register toggle information through combinational logic. To address this, we propose LEAP, the first work to enable per-cycle toggle propagation prediction with both high accuracy and efficiency. This is achieved through a novel, linear-complexity graph transformer capable of simulating toggle propagation, along with specially designed self-supervised pre-training tasks that enable the model to capture circuit structure and functionality. LEAP achieves a 7.6x speedup over the EDA tool in toggle propagation, and attains a near-perfect area under the Precision-Recall curve (PR-AUC) of 0.99 for prediction results. Moreover, LEAP can be seamlessly integrated with other machine learning based power models into LEAP-Power. This integration enables precise per-cycle layout power prediction directly from post-synthesis netlists, achieving a mean absolute percentage error(MAPE) of only 4.55%. By bypassing toggle propagation in the netlist, LEAP-Power delivers substantial runtime gains, running 5.3x faster than the model without LEAP.

cs.AR↗

arXiv2026

IMPart: Integration of Memetic Operations into Multi-Level Framework for Large-k-Way Hypergraph Partitioning

The problem of k-way hypergraph partitioning is fundamental with significant applications in various fields, including VLSI design and scientific computing. State-of-the-art hypergraph partitioners commonly employ a multi-level framework encompassing coarsening, initial partitioning, uncoarsening, and refinement phases. However, many existing methods do not scale well to problems requiring a large number of partitions (i.e., large k). In pursuit of exceptionally high solution quality, existing memetic approaches often execute their two key operations, recombination and mutation, by invoking separate, standalone multi-level partitioners. This design choice, however, renders them significantly more time-consuming than standard multi-level partitioners. To make such memetic approaches more practical, we propose an advanced memetic framework, IMPart, which introduces novel recombination and mutation operators and integrates them directly into the uncoarsening phase of a single multi-level framework. This transforms the local searches of different granularities in the traditional multi-level framework into a sophisticated, collaborative search. Experimental results on multiple standard benchmarks demonstrate our framework more effectively escapes local optima and explores the global solution space for higher-quality solutions, substantially outperforming all existing hypergraph partitioners for large-$k$-way hypergraph partitioning. Our framework highlights a new paradigm for the development of advanced hypergraph partitioners.

cs.AR↗

arXiv2026

ComPart: Community-Guided Post-Coarsening for High-Quality Hypergraph Partitioning

Hypergraph partitioning is a critical step in the design of complex embedded systems, essential for optimizing task mapping on heterogeneous MPSoCs and enabling multi-FPGA prototyping. Many existing methods rely on community detection to identify modules with dense internal and sparse external connections, typically utilizing them to constrain the coarsening phase--a widely adopted paradigm. In this work, we propose ComPart, a generalized framework that integrates diverse community detection methods to uncover high-quality clusterings throughout the post-coarsening stages (i.e., initial partitioning and uncoarsening). These discovered clusterings serve as distinct structural guides, enabling the refinement process to identify superior partitioning solutions. Our framework offers two key advantages: (1) it establishes a new paradigm that leverages community structures detected during uncoarsening to escape local optima and explore globally meaningful solution subspaces, transcending the limitations of standard local refinements; and (2) it flexibly accommodates both existing and future community detection methods. Furthermore, we theoretically generalize locally-dense decomposition--originally from graphs--to the hypergraph domain. We provide the formal extension and necessary proofs to apply this technique to hypergraphs, marking its first application in hypergraph partitioning. Specifically, we utilize this rigorously derived decomposition to guide the initial partitioning phase toward superior starting points. Experimental results on standard benchmarks demonstrate that our method consistently outperforms state-of-the-art methods in solution quality.

cs.AR↗

arXiv2026

ICP: Exploiting Instruction Correlation for Prefetching Irregular Memory Accesses

Irregular memory accesses pose challenges for effective and efficient data prefetching. While temporal prefetchers have recently shown promise for irregular memory access patterns, their effectiveness fundamentally depends on temporal address recurrence and large metadata storage. When memory addresses exhibit weak or no recurrence, as in indirect memory accesses, temporal prefetchers achieve limited performance gains while incurring substantial storage overhead. This paper proposes Instruction-Correlation Prefetching (ICP), a new hardware prefetching mechanism that exploits instruction-level correlations rather than memory-address correlations to handle irregular memory accesses. ICP observes that although memory addresses may not repeat, the instructions generating them often recur with stable data-dependency relationships. By learning these persistent instruction correlations, ICP speculatively computes and prefetches future irregular accesses using the execution results of their correlated predecessors. Across irregular SPEC CPU and GAP benchmarks, ICP outperforms the state-of-the-art temporal prefetcher Triangel by 14.0% and the indirect prefetcher DMP by 6.0%, while requiring only 2.1 KB of hardware storage, over three orders of magnitude smaller than temporal prefetchers.

cs.AR↗

arXiv2025

ReadyPower: A Reliable, Interpretable, and Handy Architectural Power Model Based on Analytical Framework

Power is a primary objective in modern processor design, requiring accurate yet efficient power modeling techniques. Architecture-level power models are necessary for early power optimization and design space exploration. However, classical analytical architecture-level power models (e.g., McPAT) suffer from significant inaccuracies. Emerging machine learning (ML)-based power models, despite their superior accuracy in research papers, are not widely adopted in the industry. In this work, we point out three inherent limitations of ML-based power models: unreliability, limited interpretability, and difficulty in usage. This work proposes a new analytical power modeling framework named ReadyPower, which is ready-for-use by being reliable, interpretable, and handy. We observe that the root cause of the low accuracy of classical analytical power models is the discrepancies between the real processor implementation and the processor's analytical model. To bridge the discrepancies, we introduce architecture-level, implementation-level, and technology-level parameters into the widely adopted McPAT analytical model to build ReadyPower. The parameters at three different levels are decided in different ways. In our experiment, averaged across different training scenarios, ReadyPower achieves >20% lower mean absolute percentage error (MAPE) and >0.2 higher correlation coefficient R compared with the ML-based baselines, on both BOOM and XiangShan CPU architectures.baselines, on both BOOM and XiangShan CPU architectures.

cs.AR↗

arXiv2025

ArchPower: Dataset for Architecture-Level Power Modeling of Modern CPU Design

Power is the primary design objective of large-scale integrated circuits (ICs), especially for complex modern processors (i.e., CPUs). Accurate CPU power evaluation requires designers to go through the whole time-consuming IC implementation process, easily taking months. At the early design stage (e.g., architecture-level), classical power models are notoriously inaccurate. Recently, ML-based architecture-level power models have been proposed to boost accuracy, but the data availability is a severe challenge. Currently, there is no open-source dataset for this important ML application. A typical dataset generation process involves correct CPU design implementation and repetitive execution of power simulation flows, requiring significant design expertise, engineering effort, and execution time. Even private in-house datasets often fail to reflect realistic CPU design scenarios. In this work, we propose ArchPower, the first open-source dataset for architecture-level processor power modeling. We go through complex and realistic design flows to collect the CPU architectural information as features and the ground-truth simulated power as labels. Our dataset includes 200 CPU data samples, collected from 25 different CPU configurations when executing 8 different workloads. There are more than 100 architectural features in each data sample, including both hardware and event parameters. The label of each sample provides fine-grained power information, including the total design power and the power for each of the 11 components. Each power value is further decomposed into four fine-grained power groups: combinational logic power, sequential logic power, memory power, and clock power. ArchPower is available at https://github.com/hkust-zhiyao/ArchPower.

cs.AR↗

arXiv2025

AutoPower: Automated Few-Shot Architecture-Level Power Modeling by Power Group Decoupling

Power efficiency is a critical design objective in modern CPU design. Architects need a fast yet accurate architecture-level power evaluation tool to perform early-stage power estimation. However, traditional analytical architecture-level power models are inaccurate. The recently proposed machine learning (ML)-based architecture-level power model requires sufficient data from known configurations for training, making it unrealistic. In this work, we propose AutoPower targeting fully automated architecture-level power modeling with limited known design configurations. We have two key observations: (1) The clock and SRAM dominate the power consumption of the processor, and (2) The clock and SRAM power correlate with structural information available at the architecture level. Based on these two observations, we propose the power group decoupling in AutoPower. First, AutoPower decouples across power groups to build individual power models for each group. Second, AutoPower designs power models by further decoupling the model into multiple sub-models within each power group. In our experiments, AutoPower can achieve a low mean absolute percentage error (MAPE) of 4.36\% and a high $R^2$ of 0.96 even with only two known configurations for training. This is 5\% lower in MAPE and 0.09 higher in $R^2$ compared with McPAT-Calib, the representative ML-based power model.

cs.AR↗

arXiv2025

Profile-Guided Temporal Prefetching

Temporal prefetching shows promise for handling irregular memory access patterns, which are common in data-dependent and pointer-based data structures. Recent studies introduced on-chip metadata storage to reduce the memory traffic caused by accessing metadata from off-chip DRAM. However, existing prefetching schemes struggle to efficiently utilize the limited on-chip storage. An alternative solution, software indirect access prefetching, remains ineffective for optimizing temporal prefetching. In this work, we propose Prophet--a hardware-software co-designed framework that leverages profile-guided methods to optimize metadata storage management. Prophet profiles programs using counters instead of traces, injects hints into programs to guide metadata storage management, and dynamically tunes these hints to enable the optimized binary to adapt to different program inputs. Prophet is designed to coexist with existing hardware temporal prefetchers, delivering efficient, high-performance solutions for frequently executed workloads while preserving the original runtime scheme for less frequently executed workloads. Prophet outperforms the state-of-the-art temporal prefetcher, Triangel, by 14.23%, effectively addressing complex temporal patterns where prior profile-guided solutions fall short (only achieving 0.1% performance gain). Prophet delivers superior performance across all evaluated workload inputs, introducing negligible profiling, analysis, and instruction overhead.

cs.AR↗

arXiv2025

Integrating Prefetcher Selection with Dynamic Request Allocation Improves Prefetching Efficiency

Hardware prefetching plays a critical role in hiding the off-chip DRAM latency. The complexity of applications results in a wide variety of memory access patterns, prompting the development of numerous cache-prefetching algorithms. Consequently, commercial processors often employ a hybrid of these algorithms to enhance the overall prefetching performance. Nonetheless, since these prefetchers share hardware resources, conflicts arising from competing prefetching requests can negate the benefits of hardware prefetching. Under such circumstances, several prefetcher selection algorithms have been proposed to mitigate conflicts between prefetchers. However, these prior solutions suffer from two limitations. First, the input demand request allocation is inaccurate. Second, the prefetcher selection criteria are coarse-grained. In this paper, we address both limitations by introducing an efficient and widely applicable prefetcher selection algorithm--Alecto, which tailors the demand requests for each prefetcher. Every demand request is first sent to Alecto to identify suitable prefetchers before being routed to prefetchers for training and prefetching. Our analysis shows that Alecto is adept at not only harmonizing prefetching accuracy, coverage, and timeliness but also significantly enhancing the utilization of the prefetcher table, which is vital for temporal prefetching. Alecto outperforms the state-of-the-art RL-based prefetcher selection algorithm--Bandit by 2.76% in single-core, and 7.56% in eight-core. For memory-intensive benchmarks, Alecto outperforms Bandit by 5.25%. Alecto consistently delivers state-of-the-art performance in scheduling various types of cache prefetchers. In addition to the performance improvement, Alecto can reduce the energy consumption associated with accessing the prefetchers' table by 48%, while only adding less than 1 KB of storage overhead.

cs.AR↗

arXiv2025

OpenLLM-RTL: Open Dataset and Benchmark for LLM-Aided Design RTL Generation

The automated generation of design RTL based on large language model (LLM) and natural language instructions has demonstrated great potential in agile circuit design. However, the lack of datasets and benchmarks in the public domain prevents the development and fair evaluation of LLM solutions. This paper highlights our latest advances in open datasets and benchmarks from three perspectives: (1) RTLLM 2.0, an updated benchmark assessing LLM's capability in design RTL generation. The benchmark is augmented to 50 hand-crafted designs. Each design provides the design description, test cases, and a correct RTL code. (2) AssertEval, an open-source benchmark assessing the LLM's assertion generation capabilities for RTL verification. The benchmark includes 18 designs, each providing specification, signal definition, and correct RTL code. (3) RTLCoder-Data, an extended open-source dataset with 80K instruction-code data samples. Moreover, we propose a new verification-based method to verify the functionality correctness of training data samples. Based on this technique, we further release a dataset with 7K verified high-quality samples. These three studies are integrated into one framework, providing off-the-shelf support for the development and evaluation of LLMs for RTL code generation and verification. Finally, extensive experiments indicate that LLM performance can be boosted by enlarging the training dataset, improving data quality, and improving the training scheme.

cs.AR↗

arXiv2024

AssertLLM: Generating Hardware Verification Assertions from Design Specifications via Multi-LLMs

Assertion-based verification (ABV) is a critical method to ensure logic designs comply with their architectural specifications. ABV requires assertions, which are generally converted from specifications through human interpretation by verification engineers. Existing methods for generating assertions from specification documents are limited to sentences extracted by engineers, discouraging their practical applications. In this work, we present AssertLLM, an automatic assertion generation framework that processes complete specification documents. AssertLLM can generate assertions from both natural language and waveform diagrams in specification files. It first converts unstructured specification sentences and waveforms into structured descriptions using natural language templates. Then, a customized Large Language Model (LLM) generates the final assertions based on these descriptions. Our evaluation demonstrates that AssertLLM can generate more accurate and higher-quality assertions compared to GPT-4o and GPT-3.5.

cs.AR↗

arXiv2024

FirePower: Towards a Foundation with Generalizable Knowledge for Architecture-Level Power Modeling

Power efficiency is a critical design objective in modern processor design. A high-fidelity architecture-level power modeling method is greatly needed by CPU architects for guiding early optimizations. However, traditional architecture-level power models can not meet the accuracy requirement, largely due to the discrepancy between the power model and actual design implementation. While some machine learning (ML)-based architecture-level power modeling methods have been proposed in recent years, the data-hungry ML model training process requires sufficient similar known designs, which are unrealistic in many development scenarios. This work proposes a new power modeling solution FirePower that targets few-shot learning scenario for new target architectures. FirePower proposes multiple new policies to utilize cross-architecture knowledge. First, it develops power models at component level, and components are defined in a power-friendly manner. Second, it supports different generalization strategies for models of different components. Third, it formulates generalizable and architecture-specific design knowledge into two separate models. FirePower also supports the evaluation of the generalization quality. In our experiments, FirePower can achieve a low error percentage of 5.8% and a high correlation R of 0.98 on average only using two configurations of target architecture. This is 8.8% lower in error percentage and 0.03 higher in R compared with directly training McPAT-Calib baseline on configurations of target architecture.

cs.AR↗

arXiv2024

AssertLLM: Generating and Evaluating Hardware Verification Assertions from Design Specifications via Multi-LLMs

Assertion-based verification (ABV) is a critical method for ensuring design circuits comply with their architectural specifications, which are typically described in natural language. This process often requires human interpretation by verification engineers to convert these specifications into functional verification assertions. Existing methods for generating assertions from natural language specifications are limited to sentences extracted by engineers, discouraging its practical application. In this work, we present AssertLLM, an automatic assertion generation framework that processes complete specification files. AssertLLM breaks down the complex task into three phases, incorporating three customized Large Language Models (LLMs) for extracting structural specifications, mapping signal definitions, and generating assertions. Our evaluation of AssertLLM on a full design, encompassing 23 I/O signals, demonstrates that 89\% of the generated assertions are both syntactically and functionally accurate.

cs.AR↗

arXiv2024

SpecLLM: Exploring Generation and Review of VLSI Design Specification with Large Language Model

The development of architecture specifications is an initial and fundamental stage of the integrated circuit (IC) design process. Traditionally, architecture specifications are crafted by experienced chip architects, a process that is not only time-consuming but also error-prone. Mistakes in these specifications may significantly affect subsequent stages of chip design. Despite the presence of advanced electronic design automation (EDA) tools, effective solutions to these specification-related challenges remain scarce. Since writing architecture specifications is naturally a natural language processing (NLP) task, this paper pioneers the automation of architecture specification development with the advanced capabilities of large language models (LLMs). Leveraging our definition and dataset, we explore the application of LLMs in two key aspects of architecture specification development: (1) Generating architecture specifications, which includes both writing specifications from scratch and converting RTL code into detailed specifications. (2) Reviewing existing architecture specifications. We got promising results indicating that LLMs may revolutionize how these critical specification documents are developed in IC design nowadays. By reducing the effort required, LLMs open up new possibilities for efficiency and accuracy in this crucial aspect of chip design.

cs.AR↗