This article introduces GenPart 2.0, an enhanced version of the hypergraph partitioner GenPart. While GenPart was limited to handling only unit vertex weights, GenPart 2.0 extends capabilities to include varying vertex weights. This extension is achieved through a variational graph neural network-based generative model and new feature preparation techniques. GenPart 2.0 addresses hypergraph partitioning challenges by establishing an embedding space that adheres to normalized cut and balance constraints. The generative model in GenPart 2.0 is designed to explore various partitioning configurations within this embedding space. Further, it enhances partitioning solutions using the V-cycle method. Testing on VLSI circuit benchmarks, including ISPD98, ISPD2005, and Titan23, under various balance constraints, has demonstrated improved performance by GenPart 2.0.

This paper introduces GPart, a scalable multilevel framework for graph partitioning that integrates GNN embeddings with efficient coarsening and refinement techniques. On the Titan23 benchmarks, GPart achieves a cut size reduction of 34.13% to 42.92% over METIS and improves cut size by 9.30% on selected DIMACS benchmarks compared to G-kway. Furthermore, experiments on the Titan23 benchmarks show that GPart reduces normalized memory usage by 24.6x compared to GAP and 12.4x compared to GenPart. Unlike existing GNN-based methods, which require large hidden layers and substantial memory, GPart’s multilevel architecture reduces hidden layer sizes, significantly optimizing memory efficiency.

While GPU-based placers have demonstrated significant speed advantages over their CPU-based counterparts, hyperparameter tuning remains a bottleneck, often requiring substantial human intervention and expert knowledge. This challenge is particularly critical given the urgent need for rapid time-to-market solutions. Recently, Large Language Models (LLMs) have exhibited remarkable capabilities in zero-shot learning, context understanding, logical reasoning, and answer generation. In this work, we introduce HyperPlace, an innovative paradigm that leverages an off-the-shelf LLM to automate hyperparameter optimization using in-context learning techniques. Our approach transcends single-output black-box optimization methods by incorporating a batch optimization mechanism that evaluates multiple hyperparameter configurations simultaneously across several GPU computing platforms. We validated the effectiveness of our approach in placement quality, measured by Half-Perimeter Wire Length (HPWL), using DREAMPlace 2.0. To further demonstrate the capability of integrating our framework with other placers, we conducted additional experiments using Xplace 2.0. By employing the ISPD2005 benchmarks for our evaluation, HyperPlace enhances the placement tools with up to a 1.66% reduction in HPWL compared to their published results. Additionally, we evaluated HyperPlace on the ISPD2015 benchmarks, which incorporate fence region constraints not present in ISPD2005 benchmarks. Under these more complex constraints, HyperPlace achieves up to a 22.24% reduction in HPWL compared to the default settings of the placement tools, further demonstrating its adaptability across diverse placement scenarios and benchmark suites.

This paper introduces GenPart, a novel hypergraph partitioner that utilizes a variational graph convolutional network-based generative model to significantly enhance partitioning performance. Traditional partitioning approaches, including multi-level and spectral partitioning, often struggle to preserve the intrinsic structure of hypergraphs, resulting in suboptimal cut performance. GenPart addresses these challenges by creating a sophisticated embedding space, compliant with normalized cut and balance constraints, and further refined by incorporating the V-cycle method. Through generative modeling, GenPart explores a variety of new and diverse partitioning configurations within this embedding space, demonstrating superiority on several VLSI circuit benchmarks and notably outperforming well-established partitioners. Rigorous testing on ISPD98, Titan23, and ISPD2005 benchmarks under varied balance constraints has proven GenPart's efficacy. Notably, on the ISPD98 benchmarks, GenPart has demonstrated the best records on 16 out of 18 instances for a balance factor of 2% and on all 18 instances for a balance factor of 10%, outperforming other state-of-the-art methods such as hMETIS, SpecPart, K-SpecPart, and MedPart. These impressive gains not only confirm GenPart's effectiveness but also suggest that it may serve as a pioneering approach in hypergraph partitioning.