DiffHLS: Differential Learning for High-Level Synthesis QoR Prediction with GNNs and LLM Code Embeddings

TL;DR

DiffHLS introduces a differential learning framework combining GNNs and LLM code embeddings to efficiently predict HLS QoR, reducing synthesis costs and improving accuracy over existing methods.

Contribution

It presents a novel differential learning approach that leverages GNNs and pretrained LLM code embeddings for more accurate HLS QoR prediction.

Findings

DiffHLS achieves lower average MAPE than GNN baselines on PolyBench.

LLM code embeddings consistently improve prediction accuracy.

Scalability validated on the ForgeHLS dataset.

Abstract

High-Level Synthesis (HLS) compiles C/C++ into RTL, but exploring pragma-driven optimization choices remains expensive because each design point requires time-consuming synthesis. We propose \textbf{\DiffHLS}, a differential learning framework for HLS Quality-of-Result (QoR) prediction that learns from kernel--design pairs: a kernel baseline and a pragma-inserted design variant. \DiffHLS~encodes kernel and design intermediate-representation graphs with dedicated graph neural network (GNN) branches, and augments the delta pathway with code embeddings from a pretrained code large language model (LLM). Instead of regressing absolute targets directly, we jointly predict the kernel baseline and the design-induced delta, and compose them to obtain the design prediction. On PolyBench, \DiffHLS~attains lower average MAPE than GNN baselines under four GNN backbones, and LLM code embeddings consistently improve over a GNN-only ablation. We further validate scalability on the ForgeHLS dataset.