Compiler-based graph representations for deep learning models of code

Brauckmann, Alexander; Goens, Andrés; Ertel, Sebastian; Castrillón, Jerónimo

doi:10.1145/3377555.3377894

Cited by 49 publications

(72 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Poem-GNN and the dense network are trained together so that the graph representation is tuned for the task. Unlike [13] that is only able to model the node connectivity, we extend the GNN to model multiple edge types (e.g., control, data, jump, token sequence, etc.). This capability allows Poem to distinguish different relationships of the code, whether it is an if branch or a function call.…”

Section: Overview Of Poemmentioning

confidence: 99%

“…The first effort in this direction is the recent work presented in [13], which employs a vanilla graph neural network (GNN) to learn representations from the graph representation of the AST or the control-data flow graphs (CDFGs). This is achieved by propagating information along the graph edges defined in a graph adjacency matrix.…”

Section: Introductionmentioning

confidence: 99%

“…Intuitively, such information would be essential for characterizing the program behavior for many coderelated tasks. By ignoring the different relationships, their GNN approach gives marginal improvement or even worse performance compared to the LSTM alternatives [13].…”

Section: Introductionmentioning

confidence: 99%

“…As a departure from prior work [11][12][13], Poem uses graphs to represent both the syntactic and semantic information of programs and employes graph-based learning methods to learn to reason over multiple graph structures. With Poem, syntax information is encoded from the AST and IR nodes.…”

Section: Introductionmentioning

confidence: 99%

“…Unlike [13] that it only uses an adjacency matrix to encode the node connectivity of the AST or CDFG, we encode different node relationships, e.g., whether it is a child-parent connection on the AST or a data flow edge in the CDFG, in different matrices and relation graphs. At the core of Poem is a novel graph neural network that can learn over multiple relationships (or edge types) simultaneously.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Deep Program Structure Modeling Through Multi-Relational Graph-based Learning

Tang

Wang

et al. 2020

Proceedings of the ACM International Conference on Parallel Architectures and Compilation Techniques

View full text Add to dashboard Cite

Deep learning is emerging as a promising technique for building predictive models to support code-related tasks like performance optimization and code vulnerability detection. One of the critical aspects of building a successful predictive model is having the right representation to characterize the model input for the given task. Existing approaches in the area typically treat the program structure as a sequential sequence but fail to capitalize on the rich semantics of data and control flow information, for which graphs are a proven representation structure. We present Poem 1 , a novel framework that automatically learns useful code representations from graph-based program structures. At the core of Poem is a graph neural network (GNN) that is specially designed for capturing the syntax and semantic information from the program abstract syntax tree and the control and data flow graph. As a departure from existing GNN-based code modeling techniques, our network simultaneously learns over multiple relations of a program graph. This capability enables the learning framework to distinguish and reason about the diverse code relationships, be it a data or a control flow or any other relationships that may be important for the downstream processing task. We apply Poem to four representative tasks that require a strong ability to reason about the program structure: heterogeneous device mapping, parallel thread coarsening, loop vectorization and code vulnerability detection. We evaluate Poem on programs written in OpenCL, C, Java and Swift, and compare it against nine learningbased methods. Experimental results show that Poem consistently outperforms all competing methods across evaluation settings.

show abstract

Section: Overview Of Poemmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Deep Program Structure Modeling Through Multi-Relational Graph-based Learning

Tang

Wang

et al. 2020

Proceedings of the ACM International Conference on Parallel Architectures and Compilation Techniques

View full text Add to dashboard Cite

show abstract

Fast selection of compiler optimizations using performance prediction with graph neural networks

Rosário

Silva

Zanella

et al. 2022

Concurrency and Computation

View full text Add to dashboard Cite

Tuning application performance on modern computing infrastructures involves choices in a vast design space as modern computing architectures can have several complex structures impacting performance. Moreover, different applications use these structures in different ways, leading to a challenging performance function. Consequently, it is hard for compilers or experts to find optimal compilation parameters for an application that maximizes such performance function. One approach to tackle this problem is to evaluate many possible optimization plans and select the best among them. However, executing an application to measure its performance for every plan can be very expensive. To tackle this problem, previous work has investigated the use of Machine Learning techniques to predict the performance of the applications without executing them quickly. In this work, we evaluate the use of graph neural networks (GNN) to make fast predictions without executing the application to guide the selection of good optimization sequences. We propose a GNN architecture to make such predictions. We train and test it using 30 thousand different compilation plans applied to 300 different applications, using ARM64 and LLVM IR code representations as input. Our results indicate that the control and data flow graph can then learn features from the control and data flow graph to outperform nongraph‐aware Machine Learning models. Our GNN architecture achieved 91% accuracy in our dataset compared to 79% when using a nongraph‐aware architecture–taking only 16ms to predict a given input. If the application been optimized took an average of 10 s to execute, and we evaluated 1000 optimization sequences, it would take almost 9 h to assess all pairs, but only 16 s with our GNN .

show abstract

Assessing the Importance of Global Relationships for Source Code Analysis Using Graph Neural Networks

Romanov

Ivanov

2023

Lecture Notes in Networks and Systems

View full text Add to dashboard Cite

Compiler-based graph representations for deep learning models of code

Cited by 49 publications

References 1 publication

Deep Program Structure Modeling Through Multi-Relational Graph-based Learning

Deep Program Structure Modeling Through Multi-Relational Graph-based Learning

Fast selection of compiler optimizations using performance prediction with graph neural networks

Assessing the Importance of Global Relationships for Source Code Analysis Using Graph Neural Networks

Contact Info

Product

Resources

About