硬件设计数据集
收藏资源简介:
硬件设计数据集由佐治亚理工学院创建,包含五个硬件设计任务的数据集,涵盖了从高层次综合到物理综合的不同电路抽象级别。数据集大小从10k+到50k+节点不等,主要用于预测硬件性能和资源使用。数据集的创建过程结合了现有的研究成果,并通过预处理使其与PyTorch Geometric兼容。该数据集主要应用于硬件设计和机器学习领域,旨在通过图表示学习技术加速硬件评估和设计优化。
The hardware design dataset, developed by the Georgia Institute of Technology, encompasses datasets for five hardware design tasks covering diverse circuit abstraction levels ranging from high-level synthesis to physical synthesis. With sizes varying from 10k+ to 50k+ graph nodes, this dataset is primarily utilized for predicting hardware performance and resource utilization. Its creation process incorporates existing research outcomes, and it has been preprocessed to be compatible with PyTorch Geometric. Targeted at the fields of hardware design and machine learning, this dataset aims to accelerate hardware evaluation and design optimization through graph representation learning technologies.

- 1A Benchmark on Directed Graph Representation Learning in Hardware Designs佐治亚理工学院 · 2024年



