IR2Vec: LLVM IR based Scalable Program Embeddings

TL;DR

IR2Vec introduces a scalable, IR-based program embedding method that captures syntax and semantics, enabling faster training and improved performance in optimization tasks across multiple platforms.

Contribution

The paper presents IR2Vec, a novel IR-based embedding infrastructure with symbolic and flow-aware encodings, outperforming existing methods in speed and accuracy.

Findings

Outperforms existing methods in device mapping and thread coarsening tasks.

Enables faster training with non-sequential models.

Achieves state-of-the-art or improved results in benchmark suites.

Abstract

We propose IR2Vec, a Concise and Scalable encoding infrastructure to represent programs as a distributed embedding in continuous space. This distributed embedding is obtained by combining representation learning methods with flow information to capture the syntax as well as the semantics of the input programs. As our infrastructure is based on the Intermediate Representation (IR) of the source code, obtained embeddings are both language and machine independent. The entities of the IR are modeled as relationships, and their representations are learned to form a seed embedding vocabulary. Using this infrastructure, we propose two incremental encodings:Symbolic and Flow-Aware. Symbolic encodings are obtained from the seed embedding vocabulary, and Flow-Aware encodings are obtained by augmenting the Symbolic encodings with the flow information. We show the effectiveness of our methodology on two optimization tasks (Heterogeneous device mapping and Thread coarsening). Our way of representing the programs enables us to use non-sequential models resulting in orders of magnitude of faster training time. Both the encodings generated by IR2Vec outperform the existing methods in both the tasks, even while using simple machine learning models. In particular, our results improve or match the state-of-the-art speedup in 11/14 benchmark-suites in the device mapping task across two platforms and 53/68 benchmarks in the Thread coarsening task across four different platforms. When compared to the other methods, our embeddings are more scalable, is non-data-hungry, and has betterOut-Of-Vocabulary (OOV) characteristics.