High-Performance GPU-to-CPU Transpilation and Optimization via High-Level Parallel Constructs

Moses, William S.; Ivanov, Ivan R.; Domke, Jens; Endo, Tetsuro; Doerfert, Johannes; Zinenko, Oleksandr

doi:10.1145/3572848.3577475

Cited by 9 publications

(3 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The pre-pruning method cannot grasp the global information of the tree and has certain blindness, and this method involves difficulty to determine whether the child nodes of clipped nodes have existence value, which may cause the decision tree to stop growing prematurely, and thus it cannot obtain the optimal decision tree. In addition, the post-pruning algorithm is usually based on statistical knowledge, and some of its parameters either depend on prior statistical laws, the domain knowledge of experts, or certain assumptions, and often need to go through repeated comparative tests to obtain satisfactory results [44].…”

Section: Methodsmentioning

confidence: 99%

“…prematurely, and thus it cannot obtain the optimal decision tree. In addition, the postpruning algorithm is usually based on statistical knowledge, and some of its parameters either depend on prior statistical laws, the domain knowledge of experts, or certain assumptions, and often need to go through repeated comparative tests to obtain satisfactory results [44].…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

A Pork Price Prediction Model Based on a Combined Sparrow Search Algorithm and Classification and Regression Trees Model

Qin,

Yang,

Zhang

2023

Applied Sciences

View full text Add to dashboard Cite

The frequent fluctuation of pork prices has seriously affected the sustainable development of the pork industry. The accurate prediction of pork prices can not only help pork practitioners make scientific decisions but also help them to avoid market risks, which is the only way to promote the healthy development of the pork industry. Therefore, to improve the prediction accuracy of pork prices, this paper first combines the Sparrow Search Algorithm (SSA) and traditional machine learning model, Classification and Regression Trees (CART), to establish an SSA-CART optimization model for predicting pork prices. Secondly, based on the Sichuan pork price data during the 12th Five-Year Plan period, the linear correlation between piglet, corn, fattening pig feed, and pork price was measured using the Pearson correlation coefficient. Thirdly, the MAE fitness value was calculated by combining the validation set and training set, and the hyperparameter “MinLeafSize” was optimized via the SSA. Finally, a comparative analysis of the prediction performance of the White Shark Optimizer (WSO)-CART model, CART model, and Simulated Annealing (SA)-CART model demonstrated that the SSA-CART model has the best prediction of pork price (compared with a single decision tree, R2 increased by 9.236%), which is conducive to providing support for pork price prediction. The accurate prediction of pork prices with an optimized machine learning model is of great practical significance for stabilizing pig production, ensuring the sustainable growth of farmers’ income, and promoting sound economic development.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

A Pork Price Prediction Model Based on a Combined Sparrow Search Algorithm and Classification and Regression Trees Model

Qin,

Yang,

Zhang

2023

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…It should be noted that OP2 is distinct from a software architecture design that uses transpilers for back-end development, which focuses on facilitating the development of back-end layers using transpilers and language interoperability for shared codebase development, whereas OP2 provides source-to-source translation and compilation framework for execution on different hardware platforms and is specifically designed for unstructured meshbased applications. • High-Performance GPU-to-CPU Transpilation and Optimization via High-Level Parallel Constructs [54] This article proposes an automated method to convert programs written in one programming model, such as CUDA, to another model, such as CPU threads, using Polygeist/MLIR, which can achieve performance portability and eliminate the need for manual application portability. This framework allows compiler transformations to be applied seamlessly, including parallelism-specific optimizations.…”

mentioning

confidence: 99%

Transpiler-Based Architecture Design Model for Back-End Layers in Software Development

Bastidas Fuertes,

Pérez,

Meza

2023

Applied Sciences

View full text Add to dashboard Cite

The utilization of software architectures and designs is widespread in software development, offering conceptual frameworks to address recurring challenges. A transpiler is a tool that automatically converts source code from one high-level programming language to another, ensuring algorithmic equivalence. This study introduces an innovative software architecture design model that integrates transpilers into the back-end layer, enabling the automatic transformation of business logic and back-end components from a single source code (the coding artifact) into diverse equivalent versions using distinct programming languages (the automatically produced code). This work encompasses both abstract and detailed design aspects, covering the proposal, automated processes, layered design, development environment, nest implementations, and cross-cutting components. In addition, it defines the main target audiences, discusses pros and cons, examines their relationships with prevalent design paradigms, addresses considerations about compatibility and debugging, and emphasizes the pivotal role of the transpiler. An empirical experiment involving the practical application of this model was conducted by implementing a collaborative to-do list application. This paper comprehensively outlines the relevant methodological approach, strategic planning, precise execution, observed outcomes, and insightful reflections while underscoring the the model’s pragmatic viability and highlighting its relevance across various software development contexts. Our contribution aims to enrich the field of software architecture design by introducing a new way of designing multi-programming-language software.

show abstract

Strided Difference Bound Matrices

Pitchanathan,

Cohen,

Zinenko

et al. 2024

Computer Aided Verification

View full text Add to dashboard Cite

A wide range of symbolic analysis and optimization problems can be formalized using polyhedra. Sub-classes of polyhedra, also known as sub-polyhedral domains, are sought for their lower space and time complexity. We introduce the Strided Difference Bound Matrix (SDBM) domain, which represents a sweet spot in the context of optimizing compilers. Its expressiveness and efficient algorithms are particularly well suited to the construction of machine learning compilers. We present decision algorithms, abstract domain operators and computational complexity proofs for SDBM. We also conduct an empirical study with the MLIR compiler framework to validate the domain’s practical applicability. We characterize a sub-class of SDBMs that frequently occurs in practice, and demonstrate even faster algorithms on this sub-class.

show abstract

High-Performance GPU-to-CPU Transpilation and Optimization via High-Level Parallel Constructs

Cited by 9 publications

References 48 publications

A Pork Price Prediction Model Based on a Combined Sparrow Search Algorithm and Classification and Regression Trees Model

A Pork Price Prediction Model Based on a Combined Sparrow Search Algorithm and Classification and Regression Trees Model

Transpiler-Based Architecture Design Model for Back-End Layers in Software Development

Strided Difference Bound Matrices

Contact Info

Product

Resources

About